Progress in Aerospace Sciences 47 (2011) 217–239

Contents lists available at ScienceDirect

Progress in Aerospace Sciences

journal homepage: www.elsevier.com/locate/paerosci

Airship dynamics modeling: A literature review

Yuwen Li a,, Meyer Nahon b, Inna Sharf b
a
Department of Aerospace Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario, Canada M5B 2K3
b
Department of Mechanical Engineering, McGill University, 817 Sherbrooke Street West, Montreal, Quebec, Canada H3A 2K6

a r t i c l e in f o abstract

Available online 14 December 2010 The resurgence of airships has created a need for dynamics models and simulation capabilities adapted to
these lighter-than-air vehicles. However, the modeling techniques for airship dynamics have lagged
behind and are less systematic than those for fixed-wing aircraft. A state-of-the-art literature review is
presented on airship dynamics modeling, aiming to provide a comprehensive description of the main
problems in this area and a useful source of references for researchers and engineers interested in modern
airship applications. The references are categorized according to the major topics in this area:
aerodynamics, flight dynamics, incorporation of structural flexibility, incorporation of atmospheric
turbulence, and effects of ballonets. Relevant analytical, numerical, and semi-empirical techniques are
discussed, with a particular focus on how the main differences between lighter-than-air and heavier-
than-air aircraft have been addressed in the modeling. Directions are suggested for future research on
each of these topics.
& 2010 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2. Aerodynamics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.1. Wind-tunnel tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.2. Potential flow aerodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
2.3. Viscous effect on hull . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
2.4. Hull–fin interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
2.5. Axial drag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
2.6. CFD results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
2.7. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
3. Flight dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
3.1. Flight tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
3.2. 6-DOF nonlinear models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
3.3. Linear models and stability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
3.3.1. Linear models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
3.3.2. Stability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
3.4. Simplified nonlinear models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
3.4.1. For evaluation of steady turn characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
3.4.2. For trajectory optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
3.5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
4. Incorporation of structural flexibility. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
4.1. Research issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
4.2. Experimental studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
4.3. FEA and CFD methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
4.4. Analytical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
4.4.1. Dynamics models of maneuvering flexible airships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
4.4.2. Wrinkling criteria. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

 Corresponding author.
E-mail addresses: [email protected] (Y. Li), [email protected] (M. Nahon), [email protected] (I. Sharf).

0376-0421/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.paerosci.2010.10.001
218 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

4.5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232

5. Incorporation of atmospheric turbulence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
5.1. Experimental studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
5.2. Theoretical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
5.2.1. Energy method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
5.2.2. Slender body method. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
5.3. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
6. Incorporation of effects of ballonets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
6.1. Experimental studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
6.2. Theoretical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
6.3. Semi-empirical models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
6.4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
7. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238

1. Introduction ballonets—components to LTA aircraft that can influence their

dynamics characteristics. Third, the HTA dynamics models are
An airship is a lighter-than-air (LTA) aircraft equipped with highly reliable for airplanes because of well-developed analytical
propulsion and steering systems. Unlike conventional heavier- and numerical techniques and a great deal of constantly updated
than-air (HTA) vehicles such as fixed-wing aircraft and helicopters, empirical data. But the dynamics modeling techniques and relevant
whose lift is aerodynamically generated by the motion of an airfoil experimental studies for airships have lagged behind and are less
through the air, airships stay aloft using a light lifting gas. This systematic than those for airplanes.
distinguishing feature can provide them with long endurance, high Existing works on airship dynamics are scattered over a range of
payload-to-weight ratio, and low fuel consumption. journals, books, and technical reports, and a large number of
Airships are the first aircraft that realized mankind’s dream of conference proceedings. These articles have been published over
controlled, powered flight. In 1784, Jean-Baptiste Meusnier pro- a period of many decades, witnessing the ascent, golden age,
posed a design for an airship of ellipsoidal shape with a rudder, an decline, and recent resurgence of airships. In 1999, Khoury and
elevator, and three large airscrews, but he lacked a lightweight, Gillett edited an important book [3] for use as a guide to the
powerful engine. Henri Giffard was the first person to apply steam- development of modern airships. The book addresses a wide range
engine technology successfully to airships. He flew his airship of issues including airship design, dynamics, control, propulsion,
17 miles in 1852, with a single propeller driven by a three and piloting. Significant progress has been made in the area of
horsepower engine [1]. The golden age of airships began with airship dynamics since then. Many new modeling techniques and
the launch of the Luftschiff Zeppelin LZ1 in 1900. The Zeppelin important issues in airship dynamics are not covered in Khoury and
airships became the most famous aircraft of that time and some of Gillett’s text, such as recent CFD methods to analyze airship
these airships were used as bombers during World War I by the aerodynamics, fluid–structure interaction in airships, the incor-
German Army. The United States and Britain built several airships, poration of atmospheric turbulence, and the effects of ballonets on
such as the R-33, R-34 British airships and the American USS airship dynamics. In 2009, Liao and Pasternak [16] published a
Shenandoah (ZR-1), in the 1920s and 1930s, mostly imitating the literature review of unconventional hull designs and airship
original Zeppelin design. But the use of airships declined over time structural analysis. However, their review is focused on airship
as fixed-wing aircraft became more capable and with the occur- designs and materials but has little information on the dynamics
rence of several airship accidents, including the 1937 burning of the characteristics.
Hindenburg, the largest aircraft ever built. Although the use of To meet the requirements of modeling and simulation for the
airships has been in recession for decades, over the past few years, resurgence of airships, a state-of-the-art literature review focusing
the advancement of modern techniques, such as composite mate- on airship dynamics is presented. This article aims to provide a
rials, optimal design, computational fluid dynamics (CFD), thermal comprehensive description of the main problems in airship
modeling, vectored thrust, and automatic control, have all brought dynamics modeling and summarize the theoretical (analytical or
a resurgence of these aircraft [2,3]. A wide range of applications numerical) and semi-empirical techniques to solve these problems.
have recently been proposed for modern airships in commercial, Five topics are discussed in this article: aerodynamics, flight
scientific, and military fields, such as advertising and tourism [4], dynamics, incorporation of structural flexibility, incorporation of
surveillance [5], environmental monitoring [6–8], planetary atmospheric turbulence, and effects of ballonets. These topics cover
exploration [9,10], heavy-lift cargo transport [11], and strato- most of the problems and modeling techniques in airship dynamics
spheric observation and telecommunication relay [12–14]. reported in literature. Aerodynamics and flight dynamics are
All of the aforementioned airship applications require a com- fundamental issues in airship design, which have been investigated
mon design tool: dynamics modeling and simulation, to analyze over many decades. In Sections 2 and 3, we first review the
their flight behavior, design their control systems, and optimize aerodynamics and flight dynamics models of airships that are
their flight trajectories. The dynamics modeling of conventional based on a rigid-body assumption. Structural strength was a
HTA aircraft has been broadly studied and has become highly significant issue in the development of old airships. However,
reliable. However, for a variety of reasons, limitations exist when the influence of structural flexibility on airship aerodynamics and
applying these HTA dynamics models, such as those developed by flight dynamics has been investigated only in the past few years
Etkin [15], to airships. First, an airships’ lift generation mechanism because of the recent advances in computational capabilities. The
is different from that of HTA aircraft due to the use of light lifting fluid–structure interaction in airships is discussed in Section 4.
gas. Certain solid–fluid interaction forces can be neglected for HTA Researchers have realized for a long time that airships’ flight
aircraft, such as buoyancy and those related to the inertia of the characteristics are vulnerable to wind and gusts. However, incor-
surrounding air, but these forces become important for airships. porating atmospheric turbulence into the mathematical dynamics
Second, airships have very different shape and structure compared models of airships is not straightforward and differs from the
with HTA aircraft. For example, many modern airships have methodologies for HTA aircraft due to the existence of light lifting
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 219

Nomenclature qei generalized elastic coordinates

q0 dynamic pressure
A, B state matrix and input matrix of the linear model R,S hull cross-sectional radius and area
CDC hull cross-flow drag coefficient Re Reynolds number
CD0 hull axial drag coefficient RT turn radius
CLa lift curve slope rG , rV , rbk position vectors of CG, CV, and ballonets from the
CF skin-friction coefficient of an equivalent flat plate body-frame origin
Cwb rotation matrix from the wind-axes frame to the u0 equilibrium forward speed
body frame VB , Vb , Vg airship volume, ballonet volume, and lifting
D drag force gas volume
Dm maximum diameter of the hull v ¼ ½u,v,wT linear velocity
FA , MA added-mass force and moment vf ¼ ½uf ,vf ,wf T flow velocity at the location of the body-
FAD , MAD aerodynamic force and moment frame origin
FC , MC control force and moment vg , vw gust speed and mean wind speed
Ff , Mf force and moment due to flow motion x, u deviation of state and control vectors from equilibrium
FN , MN aerodynamic normal force and pitch moment condition
g acceleration of gravity y1 displacement of m1 , see Fig. 20(a)
H altitude of the airship a angle of attack of the airship
h1 vertical position of m1 from the ballonet CG, see ag gust angle of attack
Fig. 20(a) b sideslip angle
I second moment of area of the hull’s cross-section e longitudinal distance from the nose
J inertia tensor of the airship ev longitudinal position at which the flow ceases to be
k1 , k2 , ku added-mass factors, see Fig. 4 potential
La length of the airship Z efficiency factor for the cross-flow drag due to
lh distance from the nose to the hull–fin inter- finite length
section point Zk , ZF hull–fin influence factors in Eq. (7)
lzb , lzg vertical distances from the hull CV to the CGs of the l equivalent distance from the CV to the point at which
ballonet air and lifting gas the total side force acts
M bending moment O wave number
M,^ C^ , K^ equivalent mass, damping coefficient, and stiffness of x ¼ ½p,q,rT angular velocity
the ballonet sloshing model in Eq. (33) o1 , z1 undamped natural frequency and damping ratio of m1 ,
MAA added-mass matrix of the airship see Fig. 20(a)
m total mass of the airship Uei shape functions
mu, Ju mass and moment of inertia of the displaced air Uv spatial gradient matrix of non-uniform flow velocity,
mb , mg masses of ballonet air and lifting gas see Eq. (22)
mb0 , Ib0 mass and moment of inertia of the ballonet air that f, y roll and pitch angles
does not move relative to the hull, see Fig. 20(a) r air density
mFij added-mass terms due to the fins sL , sH longitudinal and hoop membrane stresses of the hull
mi sloshing masses of ballonet air, see Fig. 20(a) envelope
p internal pressure yw wrinkling angle of the hull envelope
Q shear force

gas. Important works on this topic have been published in the past and aerodynamics. The former is due to the static air pressure and is
decade, which are reviewed in Section 5. Ballonets are used for the independent of the motion of the vehicle, while the latter is related
control of internal pressure in modern LTA aircraft. Considering to its motion. Most of the lift on the airship is due to aerostatics.
that the ballonets can occupy a large volume and can slosh within However, since the aerostatics is fairly simple to derive [17], the
the airship, researchers have been interested in the influence of aerodynamic computation becomes the more important issue in
ballonets on the flight of LTA aircraft. This topic is particular to airship dynamics analysis.
airship dynamics and does not appear in HTA dynamics modeling.
Relevant modeling methods are discussed in Section 6. In addition,
relevant wind-tunnel and other experimental data for airships are 2.1. Wind-tunnel tests
also summarized in the corresponding sections of this article.
Before we proceed to discuss computational aerodynamic
methods for airships, we first revisit some important wind-tunnel
2. Aerodynamics test results of these aircraft, from which we deduce the main issues
in their aerodynamics modeling.
A typical airship has a large streamlined hull filled with a light Wind-tunnel tests have been critical to airship dynamics
gas, actuated by thrusters and equipped with controllable low- analysis, especially in the golden age of airships when no accurate
aspect-ratio tail fins, as shown in Fig. 1. The thrusters, sometimes aerodynamics models were available. Two review papers on airship
vectorable, are usually installed on a gondola, although a thruster is wind-tunnel tests can be found in literature: by Jones [18] who
sometimes installed on the downward vertical fin to provide summarized the aerodynamic characteristics of several British
additional yaw control. airships from experiments in the 1910s and 1920s, and by Curtiss
The flight behavior of an airship depends upon the interaction [19] which is a review paper on LTA aerodynamic experiments up
forces between the aircraft and the air, which include aerostatics to 1976. Much wind-tunnel data on scaled airship models was
220 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

Hull Tail fin Control surface

x y
o
z
Forward ballonet z
Rear ballonet
Y

X O
Gondola
Z

Fig. 1. A typical airship.

oncoming flow. Existing methods to account for the viscous effects

Table 1
Available wind-tunnel data for some airships. will be summarized in Section 2.3.
 The tail fins, not only produce extra lift force, but also provide a
LTA model Aerodynamics results stabilizing effect on the airship against the aforementioned
unstable pitch moment. This can be observed by comparing the
Steady-state Pressure Rotational
distribution derivatives normal force and pitch moment coefficients of a bare hull to
those of the whole airship (hull plus fins), such as those plotted
Akron [20,21] | | in Fig. 3 for the Akron model. Exploiting the similarities between
Shenandoah [22] | | fins and wings, the aerodynamic force on a pair of fins joined
R-101 [23,24] | | |
YEZ-2A [25] | |
together (without the hull) can be predicted using low-aspect-
Lotte [26] | | ratio wing theories, such as those in [28], and will not be
TCOM-250 [27] | repeated here. The hull–fin interaction plays an important role
on airship aerodynamics. Curtiss [19] uncovered from the data
obtained on the R-101 and Akron models that the two fins
collected decades ago in the reports of the National Advisory directly joined together produced about 30–40% less lift than
Committee for Aeronautics (NACA) in the United States and the the two fins separated by the hull. In addition, the hull equipped
Aeronautical Research Council (ARC) in Britain, in particular, for with fins also produces more lift than a bare hull, as demon-
models of the American Akron (ZRS-4) [20,21], Shenandoah (ZR-1) strated in the force distribution at the rear of the hull for the two
[22], and the British R-101 [23,24] airships. Nowadays, wind- cases in Fig. 2. The incorporation of hull–fin interaction into the
tunnel tests are still an important means to facilitating the aerodynamic computation will be discussed in Section 2.4.
development of modern airships and other LTA vehicles, such as  The hull is the main source of the drag on an airship. On most
the YEZ-2A [25] and Lotte [26] airships and the TCOM-250 aerostat airships, between 50% and 75% of the total drag is contributed by
[27]. These wind-tunnel results include steady-state aerodynamic the hull [19], and the tail fins’ drag is about 7–27% of the hull
coefficients, pressure distributions, and rotational stability deri- drag [3]. Drag prediction methods for airship hulls will be
vatives, as listed in Table 1. reviewed in Section 2.5.
The following observations can be made from airship wind-
tunnel tests, which also reveal the main issues in airship aero-
dynamics modeling:
2.2. Potential flow aerodynamics
 The hull of an airship experiences a nose-up pitch moment at
non-zero angles of attack. This can be illustrated by the normal The aerodynamic characteristics of airships were initially
aerodynamic force distribution on the bare hull of the Akron investigated in the 1920s, and many theoretical works on the
model at an angle of attack of 123 in Fig. 2, where FN is the aerodynamics of old airships are based on potential flow theory. An
aerodynamic force normal to the hull’s centerline and e denotes important example is the report by Munk [29], in which a slender
the longitudinal position from the nose, and the normal force body assumption is applied to derive the aerodynamic normal force
per unit length is normalized by the dynamic pressure q0 . The per unit length along the hull at an angle of attack a as follows:
normal force distribution at the front and at the rear both
dF N dS
contributes to this nose-up aerodynamic moment. This aero- ¼ ðk2 k1 Þq0 sin2a ð1Þ
de de
dynamic moment is unstable because it increases as the angle of
attack increases, as shown by the pitch moment coefficients of where S is the local cross-sectional area of the hull. The factor
the bare hull of the Akron model in Fig. 3(b). The pitch moment ðk2 k1 Þ accounts for the finite length of hull, where k1 and k2 are the
MN is taken about the center of volume (CV) of the airship and is added-mass factors of ellipsoids in the longitudinal and lateral
normalized by q0 and the airship volume VB . This phenomenon directions, respectively, derived by Lamb [30]. These are plotted as
of unstable pitch moment can be explained using potential flow functions of Dm =La in Fig. 4 in which Dm is the maximum cross-
theory which we will discuss in Section 2.2. sectional diameter of the hull and La is the length of the airship [31].
 Potential flow theory cannot fully capture the real aerodynamics of The resulting force distribution at the front of the hull reasonably
the bare hull because of the effects of viscosity, especially at the matches the wind-tunnel data (see Fig. 2). The normal force and
rear of the body. When the angle of attack is close to 03 , the flow is pitch moment can be obtained by integrating the force distribution
axial and mostly remains attached to the hull. As the angle of over the hull. The normal force result obtained by integrating
attack increases, flow separates at the rear, and vortex pairs are Eq. (1) is zero, because the hull has a closed planform. The resulting
shed behind the hull. At an angle of attack of 903 , cross flow over pitch moment reveals that at non-zero angles of attack, the hull
the hull is dominant, similar to a cylinder placed normal to an experiences an unstable pitch moment, which was also observed
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 221

0.3 On bare hull (potential, Munk [29])

dFN / d / q0 (m)
0.2 On bare hull (viscous, Allen [36])
On bare hull (viscous, Hopkins [37])
0.1 On bare hull (experimental [21])

0
-0.1
-0.2
0 1 2 3 4 5 6

0.3
On hull with fins (Jones [27])
dFN / d / q0 (m)

0.2 On hull with fins (Li [34])

On hull with fins (experimental [21])
0.1
0
-0.1
-0.2
0 1 2 3 4 5 6

0.5
y (m)

-0.5

0 1 2 3 4 5
 (m)

Fig. 2. Normal force per unit length on the hull of the Akron model at an angle of attack of 121.

Bare hull (potential, Munk [29]) 0.5 Bare hull (potential, Munk [29])
Bare hull (viscous, Allen [36]) Bare hull (viscous, Allen [36])
Bare hull (viscous, Hopkins [37]) 0.45 Bare hull (viscous, Hopkins [37])
0.25 Bare hull (experiment [20]) Bare hull (experiment [20])
Hull + fins (Li [34]) 0.4 Hull + fins (Li [34])
Hull + fins (Jones [27]) Hull + fins (Jones [27]
0.2 0.35 Hull + fins (experiment [20])
Hull + fins (experiment [20])
FN / q0 / VB2/3

MN / q0 / VB

0.3
0.15
0.25

0.1 0.2
0.15
0.05 0.1
0.05
0
0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
Angle of attack (deg) Angle of attack (deg)

Fig. 3. Normal force and pitch moment coefficients of the Akron model.

from the wind-tunnel tests. Similarly, at non-zero sideslip angles, force and moment, and it can be summarized in vector form as [17]
the hull experiences an unstable yaw moment. " # " #" #
FA M11 M12 v_
Eq. (1) can be used to calculate the aerodynamic force and ¼
moment on a vehicle moving in a potential fluid with steady MA M21 M22 x_
" #
translation. A complete formulation of potential flow aerody- x ðM11 v þ M12 xÞ
namics of a 6-DOF vehicle moving in an unbounded heavy fluid   ð2Þ
v ðM11 v þ M12 xÞ þ x ðM21 v þM22 xÞ
has been developed in the field of hydrodynamics for submarines
and underwater vehicles. The potential fluid is governed by the
Laplace equation with the boundary condition that the fluid where v ¼ ½u,v,wT and x ¼ ½p,q,rT denote the 3  1 linear and
remains attached to the body surface, and then the aerodynamic angular velocity vectors of the vehicle expressed in a reference
force can be derived via Kirchhoff’s equation [30,17] or Bernoulli’s frame fixed to it, respectively. The coefficient matrix in the first
equation [31]. Kirchhoff’s equation gives the total forces and term on the right hand side is the 6  6 symmetric added-mass
moments on the vehicle from the kinetic energy of the fluid, while matrix. The first term in Eq. (2) is related to the time rates of change
Bernoulli’s equation not only provides these quantities but also can of the linear and angular velocities of the vehicle, while the second
be used to calculate the pressure distribution over the vehicle. The is related to the coupling of the linear and angular velocities. In
resulting aerodynamic loading is usually called the added-mass particular, for a vehicle in steady translation (v_ ¼ x
_ ¼ x ¼ 0), only
222 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

1 calculation for the influence of viscosity for an inclined body of

revolution, Allen [36] assumed that the viscous contribution at
k2 each station on the local cross section is equal to the steady-state
0.8 drag on a section of an infinite-length circular cylinder placed
normal to the flow. By adding a term related to this cross-flow drag
Added−mass factors

to Munk’s equation (1), the normal force per unit length along the
0.6 hull is computed as [36]
dF N dS a
¼ ðk2 k1 Þq0 sin2acos þ 2q0 RZCDC sin2 a ð4Þ
0.4 k’ de de 2
where CDC is the cross-flow drag coefficient of an infinite-length
circular cylinder, Z is an efficiency factor accounting for the finite
0.2 length of the body and determined from its fineness ratio, R is the
k1 local cross-sectional radius. The first term on the right hand side
of Eq. (4) comes from Eq. (1). The multiplication by cosða=2Þ is
0 introduced because this force is directed midway between the
0 0.2 0.4 0.6 0.8 1 normal to the body centerline and the normal to the wind direction,
Dm /La as pointed out in [36]. As plotted in Fig. 2, the difference in the
normal force per unit length between Munk’s equation (1) and
Fig. 4. Added-mass factors [31].
Allen’s equation (4) is relatively small for the hull of the
Akron model.
the moment term v ðM11 vÞ appears in Eq. (2), which is the In a similar semi-empirical method by Hopkins [37], it is
aforementioned ‘‘Munk moment’’ as it can be derived from Munk’s assumed that the aerodynamic normal force on the front portion
equation (1). of the body can be calculated using a potential flow assumption
To evaluate the potential flow aerodynamics with Eq. (2), the (i.e., Eq. (1)) while the force on the remaining portion is due to the
added-mass matrix must be computed. For a body of irregular cross-flow drag. Using this approach, the normal force distribution
shape, the added-mass matrix can be obtained using numerical on a body of revolution at low angles of attack can be obtained
approaches [32]. The exact solution for the added-mass matrix of as [37]
some regular shapes can be found in [33]. In particular, the results (
for ellipsoids [30] have been commonly used to evaluate the added- dF N 2ðk2 k1 Þq0 adS
de
if 0 r e r ev
¼ ð5Þ
mass matrices of airship hulls. If the body-fixed frame is at the CV of de 2q0 RZCDC a2 if ev o e rLa
an airship, with the x axis along the centerline and pointing to the
nose, the z axis positively downward, and the positive y axis where ev is the position at which the flow ceases to be potential.
determined by the right hand rule, the added-mass matrix of an Hopkins’ model requires the estimation of ev , which is empirically
airship with the shape in Fig. 1 can be written as determined from wind-tunnel test results as [37]
2 3 ev ¼ 0:378La þ 0:527e1 ð6Þ
k1 mu 0 0 0 0 0
6 7
6 k2 mu þ mF22 0 0 0 mF26 7 where e1 denotes the position at which dS=de has a maximum
6 7
6 k2 mu þmF33 0 mF35 0 7 negative value.
6 7
MAA ¼ 6 7
6 mF44 0 0 7 Besides separating the region where the viscous effect acts and
6 7
6 symmetric kuJuþ mF55 0 7 the small angle approximation made in Eq. (5), Hopkin’s and Allen’s
4 5
kuJu þ mF66 methods are also different in the calculation of the cross-flow drag
ð3Þ coefficient CDC . In Hopkin’s method [37,38], CDC is assumed to have
a constant value of 1.2, while in Allen’s method [36], it varies
where k1 mu, k2 mu, and kuJu in the diagonal terms are due to the hull, between 0.3 and 1.2. These differences cause different normal force
and the other terms are due to the fins (denoted by subscript F). The results from the two methods.
added-mass factors k1 , k2 , and ku are plotted as functions of Dm =La in The normal force distribution and resulting normal force and
Fig. 4, and mu and Ju are the mass and the moment of inertia of the air pitch moment for the Akron model’s bare hull from these two
displaced by the hull, respectively. The tail fins should be con- methods are compared to wind-tunnel data in Figs. 2 and 3. Note
sidered in the calculation of added-mass matrix of an airship, that the force per unit length obtained from Eq. (5) is discontinuous
because they can contribute close to 40% of the total added at ev . As illustrated in Fig. 3, the force and moment from Eq. (5) are
moments of inertia through the terms mF55 and mF66 [32,34]. in better agreement with experiment than Eq. (4) for airship hulls
Readers are referred to [34,33] for prediction methods to determine at low angles of attack. The results from Eq. (5) were validated with
the added-mass terms of the fins. wind-tunnel tests on a series of hull-shaped bodies at angles of
The added-mass matrix in Eq. (3) is given in a specific body attack up to 203 , while Eq. (4) was validated for long space-shuttle-
frame. If the body frame is located at a different position or with like bodies with a blunt base. However, Eq. (4) has the advantage
different orientation, the added-mass matrix can be obtained that it can be easily extended to the full range of angle of attack
through a transformation derived from the invariance of the kinetic from 03 to 1803 [39,40]. This extension to high angles of attack is
energy of the fluid [35]. To be consistent with the added-mass problematic for Eq. (5) due to the absence of wind-tunnel data from
aerodynamics, we also express other aerodynamic forces in the which to obtain ev at high angles of attack.
body-fixed frame in Sections 2.3– 2.5. Eqs. (4) and (5) include both the potential flow and viscous
terms, while the potential flow aerodynamics is also incorporated
2.3. Viscous effect on hull in Eq. (2). Thus, care should be taken not to overcount the potential
flow aerodynamics when combining Eqs. (2) and (4) or (5) in the
Semi-empirical methods are still used to account for the effect aerodynamic calculation. For example, if Eq. (4) or (5) is used for the
of viscosity in the computation of normal force on the airship hull steady-state pitch moment calculation, the term v ðM11 vÞ should
due to their simple formulations. To correct the normal force be dropped from Eq. (2).
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 223

2.4. Hull–fin interaction the aerodynamics after the hull–fin intersection point can be
modeled with two point forces. It is not surprising that Jones
The hull–fin interaction, as observed in wind-tunnel tests, was and DeLaurier’s model leads to more accurate force and moment
difficult to estimate in the early years of airship development. results (see Fig. 3), because the wind-tunnel data of the Akron
Prediction methods for hull–fin interaction were later investigated model were used to obtain the hull–fin influence factors Zk and ZF
for missiles and airplanes, for example, by Pitts et al. [41]. However, in their model.
these methods were not applied to airships because their devel- The extension of steady-state aerodynamic model to an airship
opment had declined by that time. in general rigid-body motion (translation plus rotation) is also
In the 1980s, Jones and DeLaurier [27] proposed a semi- discussed in [27,34]. This issue is important for the simulation of
empirical approach to predict the steady-state aerodynamics of airships in curvilinear flight. The formulation for the added-mass
aerostats1 and airships, with a particular focus on the hull–fin terms in Eq. (2) is, in fact, for general rigid-body motion, but the
interaction. In their steady-state model, an airship is divided into viscous effects in Eqs. (4) and (5) and the force in Eq. (7) presume
two distinct aerodynamic regions: the hull, extending from the steady translation. To extend the steady-state aerodynamics, Jones
nose to the hull–fin intersection point, and the fins, from that point and DeLaurier [27] discretized the airship into slices and used local
to their trailing edge. The total aerodynamic force is obtained by accelerations and velocities to calculate the aerodynamic force and
adding the contributions of the two regions. For example, the moment on different slices. They demonstrated that the trend of
normal force on an airship at an angle of attack a is computed the calculated rotational aerodynamic derivatives for the TCOM
as [27] 250 aerostat is in agreement with wind-tunnel measurements.
It should be noted that in their slice model, the aerodynamics
Zlh Zlh
a dS related to acceleration is considered, i.e., the added-mass terms are
FN ¼ ðk2 k1 ÞZk q0 sin2acos de þ q0 CDC sinasinjaj 2R de
2 de incorporated. In Li and Nahon’s paper [34], Eqs. (2) and (5), the fin
0 0 forces, and the hull–fin interaction effects are unified. To account
sin2a for the rotation of the airship, the local velocities at ev are used to
þ q0 SF CnaF ZF þ q0 SF CDCF sinasinjaj ð7Þ
2 calculate the viscous effect and the local velocities at the centers of
fins’ 1/4-chords are used for fin force calculation. They implemen-
where lh is the distance from the nose to the hull–fin intersection ted their aerodynamics computational approach in the flight
point, SF and CDCF are the reference area and cross-flow drag simulation of the Skyship-500 airship. Simulation results of time
coefficient of the fins, respectively, and CnaF is the derivative of responses to control surface deflections and steady turn rates
normal force of the fins with respect to a. The normal aerodynamic match published flight test data [34].
loads on the hull (the first two terms in Eq. (7)) are modeled as
distributed forces in a similar way to Eq. (4), but with a hull–fin 2.5. Axial drag
influence factor Zk in the first term and without the efficiency factor
Z in the second term. As well, writing sinasinjaj instead of sin2 a in Axial drag is of primary importance for computing required
Eq. (7) allows us to deal with negative angles of attack. The loads on
propulsive power and fuel consumption for airships. Wind-tunnel
the fins are modeled as two point forces (the last two terms in
data shows that the hull is the main source of drag. A popular
Eq. (7)): one at the center of pressure and the other at the center of
empirical method to predict the drag coefficient of a hull as a
area. The influence factors Zk and ZF are introduced to account for
function of fineness ratio La =Dm is presented by Hoerner [44] as
the interaction between the hull and the fins; these factors are
follows:
estimated from wind-tunnel test results. "    1:2  2:7 #
In the aerodynamic computational approach by Li and Nahon La 1=3 La La
CD0 ¼ CF 4 þ6 þ24 ð9Þ
[34], the hull–fin interaction is incorporated analytically. First, the Dm Dm Dm
effect of fins on the hull force is taken into account by adding a term
due to the fin-induced downwash. Second, the influence of the hull where CD0 is the hull’s axial drag coefficient non-dimensionalized
2=3
on the fins is taken into account by using an effective local angle of by VB in which VB is the body volume, CF is the skin-friction
attack to calculate the fins’ forces, i.e., coefficient of an equivalent flat plate and can be approximately
  computed as CF ¼ 0:045Re1=6 in which Re is the Reynolds number
R2 based on the hull length La . From Eq. (9), the minimum hull drag
ae ¼ 1 þ 2 aF ð8Þ
s (about 8 CF ) occurs when the fineness ratio La =Dm ¼ 4:65 in
where s is spanwise distance from the airship centerline and aF is the accordance with the prevailing airship designs, but the change in
local geometric angle of attack of the fin. The factor ð1 þR2 =s2 Þ accounts CD0 is small in the region close to this minimum value. The drag
for the influence of the hull [41]. Using the value of ð1þ R2 =s2 Þ at the coefficient rises by only 10% as the values of La =Dm decreases from
center of 1/4-chord, the geometric angle of attack af (representing the 4.65 to 2.8 and an even smaller drag rise is incurred for higher
lift on two fins directly joined together) is lower than the effective one values of La =Dm up to 8 [45]. The optimization of hull shape usually
ae (representing the lift on two fins mounted on a hull) by approxi- requires more accurate drag calculation using CFD techniques,
mately 36% for the Akron airship, which is in agreement with the wind- which will be discussed in the next subsection. Readers are referred
tunnel test result discussed in Section 2.1. to [3,45] for more information about the drag of other components
The normal force distribution and the resulting normal force of an airship and the drag of a spheroid-cone shaped airship.
and pitch moment are computed for the Akron model using the
methods in [27,34] to account for the hull–fin interaction and are 2.6. CFD results
shown in Figs. 2 and 3. Li and Nahon’s method [34] provides more
detail on the force distribution at the rear of the hull (see the second Recently, with the development of powerful computers, CFD
plot in Fig. 2), since Jones and DeLaurier’s method [27] assumes that techniques have been applied to analyze the aerodynamic char-
acteristics of airships. An important example is the CFD packages
1
Some references on tethered aerostats are included in this review, because
developed by Lutz et al. for the Lotte airship [32]. Their codes
they are also streamlined LTA aircraft. The incorporation of tether dynamics is couple panel methods with integral boundary-layer procedures
beyond the scope of this article and more information can be found in [42,43]. to solve the Reynolds averaged Navier–Stokes equation [32,46].
224 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

The inviscid velocity distribution resulting from the panel-method location of the separation line, but incur much higher computa-
is used as input to the boundary-layer calculation with a 3D tional costs. The paper by Wong et al. [55] implies that simple
integration procedure. The shear-layer separation line on the body computational expressions for airship aerodynamics can be
surface is obtained from the boundary-layer calculation and then obtained based on CFD results. They used source panels to model
used to improve the panel calculation by taking into account the the hull and vortex-lattice panels for the fins to calculate the
wake influence due to the shear-layer separation. This cycle aerodynamic force and moment for several airships. It is demon-
continues until a convergence of the separation line location. Lutz strated that the hull–fin influence factors Zk and Zf in Eq. (7) can be
et al. used their codes to investigate the steady-state aerodynamic obtained from the CFD results. As can be seen in Fig. 5, the trends
coefficients [47], added mass and moment of inertia [32], location exhibited by these factors from CFD results [55] are similar to those
of separation line of the hull [46,48], and propeller influence [49,50] from wind-tunnel data [27].
for the Lotte airship. Finally, it is noted that coupling CFD and FEA (finite element
CFD techniques are also used in the design of airships, in analysis) codes is a viable approach to the aeroelastic analysis of
particular, for hull shape optimization. Lutz and Wagner [51] airships and such works will be reviewed in Section 4.
and Nejati and Matsuuchi [52] developed numerical tools for the
optimization of airship hull shapes at zero angle of attack, with an
objective to minimize the drag on a hull envelope of a given volume
for a prescribed speed range. In both works, the panel codes are 2.7. Summary
coupled with an integral boundary-layer procedure to account for
the viscous effect. A source distribution on the body axis is chosen Depending on the analysis objectives, one requires different
to model the body profile and the source strengths are used as information from the aerodynamic computation. A flight dynamics
design variables for the optimization process. The main difference simulation based on a 6-DOF rigid-body model requires the total
between the techniques in [51,52] is in the calculation of transition aerodynamic force and moment on the vehicle. On the other hand,
region. In particular, Lutz and Wagner [51] used a semi-empirical an analysis which takes into account the structural flexibility
criterion to determine the transition region, while in Nejati and effects (such as membrane wrinkling and fluid–structure interac-
Matsuuchi’s method [52], the transition region is considered as a tion) requires force distribution or even pressure distribution to
point at 3% of the body length. The results in both [51,52] show calculate the deformation. Correspondingly, airship aerodynamics
that the optimal hull shape and the resulting optimal drag models can be classified into the following three categories:
coefficient vary depending on the Reynolds number regimes. To Total force/moment: The methods described in Sections 2.2– 2.5,
reduce the number of design parameters for hull optimization, either theoretical (such as Eqs. (1) and (2)) or semi-empirical (such
Wang and Shan [53] expressed the hull profile as piecewise poly- as (4), (5) and (7)), can be used to calculate the total aerodynamic
nomials using eight parameters. In their panel method, the inviscid force and moment for the bare hull or the hull–fin combination.
flow is computed from a distribution of sources on the airship Notably, Eqs. (5) and (7) and the method in [34] have been validated
surface, which provides the pressure and velocity at the body using airship wind-tunnel or flight test data.
surface. The boundary layer is computed using an integral for- Force distribution: Eqs. (1), (4), (5), and (7), and the method in
mulation of thin boundary-layer theory, and a semi-empirical [34] all provide normal force per unit length. It should be noted that
method is used to locate the transition region. Commercial CFD all these formulations give similar and reasonable force distribu-
packages have also been used to calculate the drag of a hull. Kale tion predictions on the front part of the airship. However, the
et al. [54] used FLUENT TM to obtain the drag force for a multi- aerodynamic effects at the rear are much more complicated and
objective optimization of aerostat hulls. They defined a generic accurate force distribution is still difficult to predict with these
envelope hull profile in terms of a combination of two cubic- methods.
splines, with a spherical cap in the front portion, and a parabolic Pressure distribution: The methods in Sections 2.2– 2.5 cannot
shape at the rear. The drag of the hull was reduced by 15.8% through provide the pressure distribution on the airship envelope. Readers
their optimization. are referred to [30,31] for the pressure distribution on an ellipsoid
Compared with the analytical or semi-empirical aerodynamic moving in a potential fluid, which can be obtained from the
models in Sections 2.2–2.4, CFD packages can provide more analytical solution of the Laplace equation in terms of ellipsoidal
detailed information, such as the pressure distribution and the harmonics. CFD techniques are required to compute the pressure

Fig. 5. Efficiency factors (solid line: experimental [27]; dashed line: CFD [55]).
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 225

distribution when considering viscosity, since no analytical solu- accurate than full-scale measurements by means of decelera-
tion of the Navier–Stokes equations is available for airships. tion tests.’’ [19].
The following suggestions can be made for future research on  Evaluation of airships’ lateral maneuverability consisted of turn
the development of effective aerodynamic models for airships. trials to determine their steady turn rate (yaw rate), radius RT ,
First, efforts should be made to develop and validate aerodynamic and sideslip angle b for different rudder inputs. The test data
prediction methods for airships at high angles of attack, which are showed that the yaw rate is a nonlinear function of the rudder
particularly relevant for low-speed and hovering flight. Second, input (see Fig. 6(a) for the turn rate results of the Skyship-500
given the lack of a large data base of test results on airships, CFD [58], for example), and that the product of turn radius and
packages can be used to produce the required data to extend sideslip, RT b, is approximately constant for different rudder
existing or to develop new aerodynamic computational formula- angles. Non-dimensionalized values of RT b obtained from
tions, as indicated by [55]. experiments on some old British airships [23] are plotted in
Fig. 6(b), where l is an equivalent distance from the CV to the
point of application of the total side force. More comments on
steady turn flight will follow in Section 3.4.1.
3. Flight dynamics  The flight test data was also used to identify the aerodynamic
derivatives in linearized dynamics models of airships. This will
3.1. Flight tests be further discussed in Section 3.3.

We begin this section with a brief review of airship flight test In the following subsections, we first discuss methods to
results. Reports on full-scale flight tests of airships are scarce in develop nonlinear and linear flight dynamics models of airships.
literature. During the golden age of airships, flight tests were
performed to determine the drag force and turn radius of rigid
3.2. 6-DOF nonlinear models
airships, such as the British R-29 [56] and the American USS Los
Angeles airships [57]. However, these experiments were conducted
decades ago and the accuracy of experimental results was limited A number of nonlinear flight dynamics models and simulation
by the instrumentation and measurement techniques of that time. programs have been presented for airships. In these models, the
More advanced instrumentation has been used in the flight tests airship is modeled as a rigid body with three translational and three
of modern airships, giving more accurate measurement results. rotational DOFs and hence, the dynamics model is composed of six
For example, flight tests were performed on Skyship-500 airship in nonlinear differential equations describing the translational and
the Patrol Airship Concept Evaluation (PACE) program [58,59] rotational dynamics. The dynamics equations of motion for an
in the 1980s and on the Lotte airship more recently [60]. In these airship written in scalar form can be found in [25]. In matrix/vector
tests, the responses to inputs of elevator, rudder, and throttle were form, they can be summarized as
" #" # " #
measured. mI33 mr G v_ mx vmx r Gx
We can summarize the flight test results by grouping them into þ
mrG J x_ mr  
G x v þ x Jx
three categories. " # " # " # " #
mg rVB g FAD FC
¼ 
mrGg rVB rV g þ MAD þ MC ð10Þ
 Deceleration tests conducted in 1920s and 1930s were used to
obtain the drag of old airships by measuring the velocity
response during periods of zero thrust. The drag force was then where m is the total mass, including the hull, gas, gondola, fins,
deduced by using a 1-DOF model of an airship in straight, level, ballonets, etc., J is the inertia tensor; rG and rV are the position
and neutral buoyancy flight [45]. However, a number of vectors of the center of gravity (CG) and the CV expressed in the
problems plagued this approach [19]. First, it was difficult to body frame, respectively; and g is the acceleration of gravity.
maintain a constant altitude in the deceleration flight and, Vectors FAD , MAD and FC , MC denote the aerodynamic and control
consequently, elevator inputs and pitch rotation contaminated force and moment, respectively. Recall that v and x denote the
the data. Second, measurement time lags were not taken into linear and angular velocity vectors of the airship-fixed frame.
account in the data processing. As a result, ‘‘airship zero-lift drag The left hand side of Eq. (10) represents the inertial forces on the
prediction based on a proper combination of scale model tests airship, with the first term related to the acceleration and the
and the theoretical/empirical correction was (as of 1935) more second term representing the nonlinear centrifugal and Coriolis

10 2
2−DOF model [66]
6−DOF model [34]
5 experiment [58] 1.5
Yaw rate (deg/s)

0 1

−5 0.5 RT β / λ from experimental data

1 / (k2 − k1)
−10 0
30 15 0 −15 −30 0 5 10 15
Rudder deflection (deg) Lm/Dh

Fig. 6. Steady turn results: (a) turn rates of Skyship-500 at 12.86 m/s and (b) RT b=l of some old British airships [23].
226 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

terms. The inertial parameters of the airship are required in interpolation and extrapolation of the discrete data points to
Eq. (10). For a neutrally buoyant condition, the mass must be calculate the aerodynamic forces and moments.
m ¼ rVB ; however, determination of the CG’s position vector rG and Flight dynamics of unconventional airships have also been studied.
the inertia tensor J require details on the design of the airship. It will NASA jointly with Systems Technology, Inc. performed a feasibility
be demonstrated in Section 3.4.1 that a simplified model can be study of a hybrid heavy-lift airship, which consisted of a helium-filled
used to predict steady turn rate, without the use of rG and J. envelope mounted on a frame with a helicopter at each corner, as
The right hand side of Eq. (10) includes the external forces and displayed in Fig. 7(c). Ringland et al. [63] derived the nonlinear 6-DOF
moments due to gravitational, aerostatic, aerodynamic, and control equations of motion and developed the simulation program HLASIM to
forces, respectively. In some models, the added-mass effects, evaluate this airship’s flight characteristics, with a particular focus on
incorporated here through FAD and MAD , are treated as virtual the influence of high incidence and atmospheric turbulence on the
inertia and are incorporated on the left hand side of Eq. (10); either flight response. It is noted that this type of hybrid airship has very
approach must produce the same numerical results [61]. The different flight characteristics from conventional airships due to the
control forces are due to thrusters and deflection the control aerodynamic lift forces generated by the helicopters.
surfaces. Prediction of control surface forces has been investigated
thoroughly in the context of conventional aircraft, as for example, 3.3. Linear models and stability analysis
in [28], and these results will not be repeated here. The dynamics of
thrusters is seldom discussed in the literature on airships. 3.3.1. Linear models
Eq. (10) is written in the body-fixed frame at an arbitrary The equations of motion (10) can be linearized with the
location, and the usual choice for the body frame’s origin is different assumption that the motion of the aircraft is constrained to small
for airships and airplanes. Usually, for airplanes, the body-fixed perturbations about a trimmed equilibrium flight condition. Linear
frame is located at the CG (i.e., rG ¼ 0) so that the centrifugal terms dynamics models have been widely applied to study the flight
are eliminated and the translational and rotational accelerations behavior of airships, as for example in [64], and for control design,
are decoupled in the equations of motion. However, for airships, it as is done in [12] for a large high-altitude airship. The linearized
is more convenient to establish the body-fixed frame at the CV equations of motion can be divided into the longitudinal and
(rV ¼ 0) for two reasons: (1) the location of the CG can vary lateral-direction sets, either of which can be written as
substantially during flight whereas the CV remain fixed, and
mx_ ¼ ax þ bu or x_ ¼ Ax þ Bu ð11Þ
(2) the computation of the added-mass matrix is simplified.
Flight dynamics models have been developed for several air- where x and u are the deviations of state and control vectors from the
ships. One of the differences between these models is their equilibrium condition, and A ¼ m1 a, B ¼ m1 b are state and control
approach to compute the aerodynamics. Amann [62] adopted matrices. The state and control vectors can be defined as x ¼ ½u,w,q, yT
the aerodynamics model of Jones and DeLaurier [27] in a nonlinear and u ¼ ½T, de T for the longitudinal motion, and x ¼ ½v,p,r, fT and
dynamics simulation program to predict the time responses due to u ¼ dr for the lateral motion, where y and f are pitch and roll angles of
different control inputs for the Skyship-500 airship (see Fig. 7(a)). the body-fixed frame, de and dr are the elevator and rudder deflections,
Li and Nahon [34] unified a series of aerodynamic computational and T is the thrust force. Matrix m is the mass matrix, and a and b
methods for the nonlinear dynamics simulation of the same contain the derivatives of the external force or moment with respect to
airship. Wind-tunnel data can also be used to determine the the corresponding states (refer to [3] for detailed expressions).
aerodynamic characteristics needed in nonlinear flight dynamics In formulating linear models of the Lotte airship (see Fig. 7(d)),
simulation. For example, Gomes [25] imported the wind-tunnel Kornienko [60] found that some derivatives in a and b were
aerodynamic coefficient results into their ACSL nonlinear simula- insignificant and could be ignored, as listed in Table 2. For example,
tion program for the YEZ-2A airship (see Fig. 7(b)) and used since the thruster of the Lotte airship is mounted at the rear of the

Fig. 7. Some simulated airships: (a) Skyship-500, (b) YEZ-2A [25], (c) hybrid heavy lift airship [63] and (d) Lotte [26].
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 227

hull, a thrust force perturbation only has a small influence on the motion, similar to a pendulum. Stability has been analyzed for
pitch moment, yielding @My =@T  0. However, this is not true if the airships using analytical linear models [25,60,64] or numerical
thruster is mounted on the gondola, well below the CV, and models obtained from finite differencing of the nonlinear equations
therefore this derivative can be important for other airships. [66]. These stability studies are carried out by evaluating the
Linear models can be calibrated by using wind-tunnel or flight eigenvalues and eigenvectors of the state matrices of linear models.
test data. Jex and Gelhausen [59] adopted the HLASIM simulation Stability analysis results in [25,60,64,66] reveal that conventional
program to calculate the aerodynamic derivatives in the linear airships have similar motion characteristics for each mode. In the
dynamics model of the Skyship-500, and the matrices A and B were longitudinal plane, the first mode is a surge subsidence mode caused by
improved by using a frequency-domain fitting technique on the aerodynamic axial drag, and can be considered as a 1-DOF surge motion
flight test data. Yamasaki and Goto [65] carried out a series of flight u. The second longitudinal mode is a heave-pitch subsidence mode
tests for the identification of matrices m, a, and b in the linear caused by the normal aerodynamic drag. The dominant motion is the
models for a small indoor airship, and the test data were analyzed heave motion w near zero speed, coupling with some pitch angle y and
with an extended least-squares algorithm to obtain the added mass pitch rate q as the speed increases. The third longitudinal model is a
and aerodynamic stability derivatives. Another model identifica- pitch-incidence oscillatory mode caused by the fact that the CG of an
tion study of airship dynamics was conducted by Kornienko [60]. airship is located below the CV. Near zero speed, the dominant motion
The responses of the Lotte airship due to inputs of throttle, is the pitch rate q and as the speed increases, w becomes apparent. In
elevators, rudders, and ailerons were measured and the flight test the lateral plane, the first mode is a sideslip-yaw subsidence mode.
data was used to determine the aerodynamic derivatives with Near zero speed, the most apparent motion is sideslip v and yaw rate r
output-error and filter-error methods. and with increasing speed, the roll rotation f becomes apparent
because of the centrifugal force. The second lateral mode is a yaw-roll
subsidence mode, which is coupled with the first mode. The yaw rate
3.3.2. Stability analysis and sideslip are dominant at near-zero speed and they couple with the
Flight stability analysis is important for aircraft performance roll rate p as the speed increases. The third lateral mode is a roll
evaluation. Although most of the airship lift is generated by the oscillation due to the offset of the CG from the CV. The readers are
aerostatic forces, the aerodynamic characteristics determine the referred to [25,60,66] for detailed eigen diagrams for conventional
stability of the aircraft. On the one hand, the streamlined hull airships at different speeds.
experiences the aforementioned unstable pitch and yaw Munk Only two modes can become unstable according to the stability
moments; on the other hand, the viscous effects acting at the rear of analysis results in [25,60,64,66]. As the speed increases, the pitch-
the hull and the aerodynamic forces acting on the fins tend to incidence oscillation mode and the sideslip-yaw subsidence mode
stabilize the system. In addition, since the CG of an airship is located of the Lotte airship become unstable, as illustrated by the eigen-
below the CV, the aircraft experiences oscillatory pitch and roll values at equilibrium forward speeds u0 of 3, 4, 8, and 12 m/s in
Fig. 8. This is because the unstable Munk moment dominates over
Table 2 the stabilizing moment generated by the fins [60]. All other modes
Insignificant derivatives in the linear models of the Lotte [60]. become more stable as the speed increases, due to the increasing
aerodynamic damping.
Forces and moments Insignificant derivatives

Axial force Fx @Fx =@w, @Fx =@q, @Fx =@de

Vertical force Fz @Fz =@u, @Fz =@y, @Fz =@T 3.4. Simplified nonlinear models
Pitch moment My @My =@u, @My =@T
Side force Fy @Fy =@p 3.4.1. For evaluation of steady turn characteristics
Roll moment Mx @Mx =@v, @Mx =@r
Yaw moment Mz
Steady turn characteristics are a key aspect in the evaluation of
@Mz =@p, @Mx =@f
the lateral maneuverability of airships. A turn can be initiated by

Fig. 8. Eigenvalues of the linear model of Lotte airship [60].

228 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

Fig. 9. An airship in a steady turn.

applying a rudder deflection input, as shown in Fig. 9. The airship with the lift and drag forces obtained from wind-tunnel tests. This
with a forward speed u is given a rudder deflection dr with the point-mass model is then used to obtain the ascent trajectories
trailing edge left, which generates a negative yaw moment and a with minimum time or minimum energy. Mueller et al. [68] used a
positive lateral velocity v. This lateral velocity leads to a negative dynamics model similar to that in [13] also for the ascent trajectory
Munk yaw moment and the airship starts to turn with a negative optimization of stratospheric airships. They further refined the
yaw rate. This yaw rate and lateral velocity increase the aero- dynamics model to account for the earth’s rotation and the
dynamic force on the vertical fins resulting in a positive yaw deflation of ballonets.
moment. Once the moment from the rudder, the Munk moment, The simplification of a point-mass model is adequate for the
and the positive aerodynamic moment due to the fins cancel each applications considered in [13,68] because the flight path angle,
other, the airship is in a steady turn. heading angle, and bank angle are slowly varying so that the
Simplified models have been applied to predict the steady turn rotational dynamics of airships can be ignored. Unlike in the previous
characteristics, and these require much less information than a full papers reviewed here, these models are established in a wind-axes
6-DOF flight simulation. Jones [23] derived a simplified 2-DOF model to local frame foxw yw zw g, whose orientation is different from the body-
investigate the steady turn characteristics for the R-101 airship, with fixed frame in Fig. 1. The origin of the wind-axes frame is also on the
the aerodynamic derivatives measured from wind-tunnel experiments. airship but its xw axis is aligned with the airship’s speed relative to the
Li [66] used a similar 2-DOF dynamics model to predict the steady turn wind rather than along the centerline. The use of the wind-axes frame
rates for the Skyship-500. The result from the 2-DOF model was in to define the airship dynamics is consistent with normal practice in
agreement with the results from the full 6-DOF model and the flight trajectory planning for airplanes [69]. However, care should be taken
tests (see Fig. 6(a)). The implementation of these simplified 2-DOF in the use of the wind-axes frame for airships because of potential
models requires only the dimensional information about the airship difficulties in accounting for the added-mass effects. As pointed out in
and does not require the position of CG and the inertia tensor. In [70], the added-mass matrix M11 in Eq. (2), which is expressed in the
addition, Jones [23] determined that the product of the turn rate and body-fixed frame oxyz in Fig. 1, should be converted to the wind-axes
the sideslip RT b is approximately constant for different rudder frame using [70]
deflections from the side force and yaw moment balance of an airship
in a steady turn, i.e., M11,w ¼ CTwb M11 Cwb ð13Þ

RT b 1 where Cwb is the rotation matrix from the wind-axes frame to the
¼ ð12Þ
l k2 k1 body frame foxyzg.
Recall that l is an equivalent distance from the CV to the point at which
the total side force acts. The curve of 1=ðk2 k1 Þ is plotted in Fig. 6(b) as
a function of fineness ratio La =Dm . It should be noted that the 3.5. Summary
aforementioned 2-DOF models and Eq. (12) can be used only for
airships without auxiliary lateral thrusters, because the side force and The main issues in the flight dynamics modeling of airships are the
yaw moment balance equations will change if auxiliary lateral prediction of aerodynamic force and moment, as discussed in detail in
thrusters are used. Section 2, and the inertial parameters of the airship. Once these are
properly formulated, the flight dynamics can be simulated using the
3.4.2. For trajectory optimization nonlinear motion equations (10) or linear equations (11).
Simplified nonlinear models have been used for trajectory The flight characteristics of airships are different from those of
planning of stratospheric airships in the last few years. Bestaoui airplanes due to the light lifting gas and the fact that the CG is well
and Hima [67] carried out a preliminary study on the trajectory below the CV. The stability of an airship depends on whether the fin
planning of a small blimp, using the 6-DOF nonlinear dynamics in force and the restoring gravitational moment are sufficient to
the body-fixed frame with the assumption of trimmed flight, i.e., stabilize the vehicle against the unstable Munk moment. The
v_ ¼ x
_ ¼ 0 in Eq. (10). Little information is given in [67] on the modes of airships change dramatically with speed, implying that
computation of aerodynamic coefficients. To overcome the limita- different controllers would be required for different speed regimes.
tion of trimmed flight, Lee and Bang [13] published a paper on the Steady turn behavior is also primarily affected by the aerodynamic
ascent trajectory optimization of a stratospheric airship in the characteristics. For this reason, it is possible to predict the steady
presence of atmospheric turbulence. The airship is modeled as a turn radius without using the inertial parameters. In addition to
point mass such that its rotational dynamics is ignored. The thrust, stability and lateral maneuverability analyses, flight dynamics
drag, buoyancy, and gravitation forces are incorporated into the modeling is also integral to the solution of trajectory optimization
calculation of the airspeed, flight path angle, and heading angle, problems.
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 229

4. Incorporation of structural flexibility make the fluid–structure interaction a less important issue in airship
development. However, there has been growing interest in this
4.1. Research issues problem over the last few years, partly because of the recent advances
in computational capabilities which make it possible to solve the
Unlike an airplane, for which the most significantly deformable fluid–structure interaction problems, but also because of proposed
components are the wings, the most flexible component of an airship new airship designs and materials.
is its hull. Based on the hull structure, airships fall into three main Fig. 10(a) shows the Sanswire Stratellite airship, designed by
categories, namely, rigid airships, non-rigid airships, and semi-rigid Sanswire-TAO [78] for missions in the lower stratosphere up to
airships. Rigid airships have a rigid hull frame containing multiple, 18,288 m (60,000 ft) altitude. The Stratellite is a segmented airship
non-pressurized gas cells or balloons to provide lift. Both semi-rigid consisting of a sequence of buoyancy cells. Flight tests with the
and non-rigid airships use an inflated envelope as a hull and the hull airship demonstrated large deformations of the hull which results
shape is retained by a pressure level in excess of the surrounding air from the relative displacement between the buoyancy cells. There-
pressure. In contrast to non-rigid airships, semi-rigid airships have a fore, the aerodynamics and flight dynamics models based on a
rigid keel along the bottom of the envelope to distribute suspension rigid-body assumption, such as those reviewed in Sections 2 and 3,
loads onto the envelope. Non-rigid airships are the most common are not likely to be applicable to this very flexible airship.
construction nowadays. Sufficient pressure difference between the Recently, very thin films have been proposed for high-altitude
surrounding air and the internal lifting gas is maintained across airships, for example, the High Platform II shown in Fig. 10(b), the first
altitudes by inflating or deflating ballonets, which are air bags stratospheric airship in the world [79]. These materials were originally
contained inside the hull (see Fig. 1). Unconventional hull structure developed as part of NASA’s Ultra Long Duration Balloon project and the
designs, such as the double-hull airships and winged airships, have thickness of the films can be as small as 0.038–0.048 mm [72,73].
also been proposed in the past two decades and the readers are Although they are very light and provide sufficient strength, these films
referred to [16] for more information on these designs. have lower elastic moduli and are much thinner than conventional
Structural strength was a significant issue in the development of airship envelope materials such as Nylon and Dacron. Accordingly, the
old rigid airships. For example, Evans [71] computed the force aeroelastic effects on the flight behavior could be important for the
distribution due to aerostatic, aerodynamic, and inertial forces for airships made of these unconventional composite materials.
the Shenandoah rigid airship and demonstrated the possibility of
catastrophic failure due to the structural bending moment. To 4.2. Experimental studies
avoid the structural vulnerability, rigid designs are seldom used in
modern airship construction and hence, we focus our attention on Experimental studies of airship fluid–structure interaction are
non-rigid and semi-rigid airships. very limited. Although there are no quantitative results on the
Research on modern airship structures is conducted from two structural deformation of airships (from wind-tunnel or flight tests),
perspectives: (1) material behavior and (2) fluid–structure inter- some qualitative observations can be gathered from the literature.
action. Examples in the first category include the development of Aeroelastic effects were observed for the old German Siemens-
new airship envelope materials [72,73], experimental studies on Schuckert airship in 1910s [80]. It was found that the airship was
the tear propagation of airship material [74], and the FEA stress deformed like a banana due to side forces on the vertical tail planes.
analysis for the envelopes of LTA aircraft [75]. This review focuses This indicates that bending was the main deformation of this airship.
on problems in the second category, since our interest is in airship Two German engineers, Hass and Dietzius, performed experi-
dynamics. Research in fluid–structure interaction of airships ments on a non-rigid airship model filled with water, to investigate
addresses two issues: how an airship deforms in flight and how the validity of beam theory assumptions to describe the deforma-
the deformation influences the aerodynamics and flight dynamics. tion of the airship [81]. After bending loads were applied to the
Although aeroelastic theories are well developed for airplanes [76] model, it was observed that the cross-sections remained planar
and techniques are available for modeling the mutual interaction after the deformation. The authors of [81] found that the Navier
between flexibility, aerodynamics, and flight dynamics of maneuver- hypothesis and Hooke’s law may be assumed valid for the bending
ing HTA aircraft [77], only a few investigations have been reported on of non-rigid airship envelopes. As a result, the longitudinal tension
the fluid–structure interaction of LTA aircraft. There are two reasons stress of the envelope could be calculated as [81]
for this. First, the LTA fluid–structure interaction problem is complex,
especially in light of our far from mature understanding of their Mc
sL ¼ ð14Þ
aerodynamics and structural behavior. Second, LTA aircraft fly at low I
speeds and the structural vulnerability and safety have been greatly where M is the internal bending moment, c is the distance to the
improved for modern airships made of membrane structures, which neutral axis, which is the centerline of the airship, and I is the

Fig. 10. (a) The Sanswire Stratellite airships [78] and (b) the High Platform II [79].
230 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

10
0.025
V = 21.5 m/s

Elastic displacement (m)

8 V = 33.3 m/s 0.02

6 0.015

CLα
4 0.01

2 0.005

0 0
0 50 100 150 200 250 0 0.2 0.4 0.6 0.8 1
ε (m) 1/2ρ V2 (kPa)

Fig. 11. Aeroelasticity simulation results from [82]: (a) deflection and (b) lift curve slope.

second moment of area about the neutral axis. This study indicates
that the deflection behavior of a non-rigid airship can be repre-
sented using a bending beam model. Rigid
Linear elastic
4.3. FEA and CFD methods Nonlinear elastic

CFD packages have been used jointly with FEA modeling tools to
study the aerodynamics of deformed airships. Amiryants et al. [82]
investigated the aeroelasticity characteristics of a 260-m semi-
CL

rigid airship using FEA packages NASTRAN and ABAQUS for the
structural analysis and using ARGON for the aerodynamics com-
putation. They demonstrated that the first two modes of the airship 0
could be described as beam-bending modes and, thus, a simplified
beam model could be applied to compute the deflection of the
airship. From their static aeroelastic analysis results, the airship
deflected in a banana shape and considerable elastic displacement
(maximum at the nose of approximately 3.7% of the total length)
was observed in their simulation when the airship was at a sideslip −4 −2 0 2 4
of 53 with a speed of 33.3 m/s (see Fig. 11(a)). The lift on the forward
α (deg)
part of the deformed airship increased, while that on the rear
decreased as a result of hull bending. As shown in Fig. 11(b), the Fig. 12. Lift coefficient of an elastic airship [80] (values on the vertical axis were not
resulting lift curve slope CLa decreases compared to the rigid model, provided in [80]).
for which CLa remains constant at the zero-speed value, and this
reduction becomes more obvious as the speed increases. aeroelastic analysis of several non-rigid airship hulls varying in length
Bessert and Frederich [80] analyzed the effects of deformation from 20 to 220 m. The lift coefficient results of these hulls are plotted
on the lift coefficients CL for the 260-m cargo lift CL-160 airship in Fig. 13 and they show a relatively small effect of elasticity on CL ,
using the FEA package ABAQUS and CFD solver VSAERO. The fluid– albeit increasing slightly as the airship length increases.
structure interaction is accounted for by exchanging the geometric It is difficult to draw any definitive conclusions based on the
and aerodynamic solutions at every time step of the co-simulation. above works about the influence of deformation on airship aero-
The resulting lift coefficients based on rigid-body, linear elastic, dynamics, because there is little consistency in the findings
and nonlinear elastic models are compared in Fig. 12. The zero-lift reported in different papers. Also, the simulated airships have
angles of attack from the two elastic models are different from that different physical and geometric parameters and the accuracy of
of the rigid model due to the deformations. The lift curve slope CLa the CFD and FEA packages may vary.
from the linear elastic model is close to the rigid model. However,
the nonlinear elastic model predicts a noticeably different slope.
4.4. Analytical models
Unlike [82] where the lift curve slope CLa is reduced by the
deformation, the results in [80] show that the deformed airship
A few works have been published on analytical models for the
has a larger CLa than the rigid airship.
fluid–structure interaction in airships, where researchers incorpo-
Other published results seem to indicate that the effects of
rate the structural deformation into airship flight dynamics.
structural flexibility are rather small. Omari [83] developed a
It should be noted that the effects of envelope wrinkling are not
numerical structure–fluid interaction solver based on a mixed
considered in these models. However, since wrinkling is a problem
element volume discretization. The deformation, vibration modes,
specific to in LTA aircraft and for the sake of completeness, the
and aerodynamic derivatives were calculated for an 1.37-m ellipsoi-
formulation of wrinkling criteria will also be reviewed here.
dal membrane in an inviscid flow. The computed first and third modes
of the body were beam-bending modes and the second mode was a
membrane mode. The influence of elastic deformation on the lift 4.4.1. Dynamics models of maneuvering flexible airships
coefficient was found to be small. Liu et al. [84] combined a nonlinear Venkatesan and Friedmann [85] conducted an aeroelastic analysis
FEA model and a CFD model based on finite volume method for the of the hybrid heavy lift airship, shown earlier in Fig. 7(c). The main
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 231

Fig. 13. Lift coefficients of elastic airship hulls in [84].

N
u Σ qei Φei
i=1 ΔFz
w
y
o
x
z r
Munk moment

Fz

Fig. 14. An airship in bending.

concern in their work was the fluid–structure interaction between the deflection at the rear causes a decrease in the local angles of attack.
helicopters and the supporting structure and thus it is not applicable This effectively produces a downward force DFz , which reduces the
to conventional airships. Bennaceur et al. [86] investigated the stabilizing aerodynamic effects of viscosity and fin force. It is worth
equations of motion of a flexible conventional airship with a particular noting that the simulation results for the Skyship-500 in [87]
focus on the effects of deformation on the inertial force. However, demonstrate that this deformation-induced force DFz is small for
their formulation was limited by the potential fluid assumption and this airship due to its fairly high bending stiffness.
little discussion was given on the interaction between the flexibility
and the aerodynamic forces.
Li et al. [87] proposed a theoretical framework for the dynamics 4.4.2. Wrinkling criteria
modeling of flexible airships, which integrated the flight dynamics, Wrinkling is a particular phenomenon that makes the structural
structural dynamics, aerostatics, and aerodynamics. The elastic characteristics of LTA aircraft different from those of HTA aircraft;
P by definition, it occurs when the stress of the inflated hull envelope
deformation was written as N i ¼ 1 qei Uei , in which qei are the time-
dependent generalized coordinates and Uei are the time-indepen- becomes zero. The longitudinal and hoop membrane stresses for an
dent shape functions taken as natural vibration modes of an Euler– inflated cylinder, in the absence of external loads, can be written as
Bernoulli beam, and N is the number of shape functions. The normal [89,90]
force per unit length was predicted for the calculation of bending pR pR
deflection. To capture the fluid–structure interaction, local velocity sL ¼ , sH ¼ ð16Þ
2t t
distribution on the deformed airship was used in the aerodynamics
calculation, i.e., where t is the membrane thickness, R is the radius of the cylinder,
and p is the internal pressure. As well, treating the cylinder as a
X
N X
N beam, the longitudinal stress due to the bending moment M can be
v d  v þ x r þ q_ ei Uei u qei Uuei ð15Þ calculated from Eq. (14), with c ¼ R. Wrinkling occurs when the
i¼1 i¼1
maximum compressive stress due to bending equals to the long-
where r is the position vector of a material point in the body-fixed itudinal stress sL due to internal pressure, as in Eq. (16), yielding
frame, and the last two terms represent the local velocity due to the wrinkling criterion for bending moment as [3]
deformation. The work of Li et al. [87] reveals two main results:
ppR3
first, the potential flow aerodynamics can significantly reduce the M4 ð17Þ
2
natural frequencies of an airship (by 25–30%), and second, the
bending deflection reduces the stabilizing aerodynamic effects A relationship between the wrinkling angle and the bending
acting at the rear of the airship. With respect to the natural moment of the inflated cylinder can be further derived as [3]
frequencies, similar conclusions have been reported for other
M pyw þ cosyw sinyw
vehicles moving in a heavy fluid, such as ships [88]. The second ¼ ð18Þ
ppR3 2½sinyw þ ðpyw Þcosyw 
finding mentioned above can be explained by considering the
schematic in Fig. 14, which shows a flexible airship moving with a where the wrinkling angle yw defines the wrinkled portion in a
forward speed u and a vertical downward speed w. The vertical cross section, as illustrated in Fig. 15. From that figure, one can see
speed leads to an upward aerodynamic force Fz at the rear due to that the wrinkling starts at M ¼ ppR3 =2 and as the bending moment
viscosity and fins, which tends to stabilize the airship against the increases to ppR3 , the wrinkling angle approaches p, i.e., the
Munk pitch moment. Due to the bending deflection, the airship wrinkles propagate completely around the circumference of the
deforms in a nose- and tail-up banana shape. The upward cylinder.
232 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

1 deformation have on the lift coefficient and the lift-curve slope? How
do these effects scale with airship size? Furthermore, predictions from
different FEA and CFD packages need to be compared by employing a
0.8 set of benchmark tests.
The following suggestions can be made for future research on
fluid–structure interaction in airships. First, efforts should be made
0.6
to derive simple analytical aeroelastic models for airships, for
Wrinkling
pR3

w example, to predict the fin force reduction due to bending deflec-

M

0.4 tion as indicated in [87]. Second, prediction methods should be

developed for the influence of wrinkling on the aerodynamics and
M M
flight dynamics. Third, efforts should be made to investigate the
0.2 aeroelastic characteristics of unconventional airship designs made
of very thin films or with large deformation.

0
0 0.5 1 1.5 2 2.5 3 5. Incorporation of atmospheric turbulence
w
As with LTA aircraft, an airship’s flight characteristics can be
Fig. 15. Relationship of wrinkling angle and bending moment from Eq. (18). strongly affected by atmospheric turbulence. In this section, we
focus on how to incorporate a particular representation of atmo-
A wrinkling criterion for the internal shear force is also derived spheric turbulence into the airship dynamics model. The issue of
in [89,90]. In shear, wrinkling occurs if the minimum principal how to obtain the atmospheric turbulence (either by measure-
membrane stress becomes zero. Under this condition, the max- ments or modeling) is beyond the scope of this review. Readers are
imum allowable shear stress can be determined for the hoop and referred to [91] for a detailed presentation on the modeling of
longitudinal stresses in Eq. (16). The resulting wrinkling criterion is atmospheric turbulence.
as follows:
5.1. Experimental studies
ppR2
Q 4 pffiffiffi ð19Þ
2 Quantifying the motion response of an airship to the atmo-
spheric turbulence through flight tests is a difficult task, since this
where Q is the shear force.
requires simultaneous measurements of airship motion and tur-
The wrinkling criteria in Eqs. (18) and (19) are based on the
bulent wind. Little data on atmospheric turbulence and the
stress analysis of an inflated cylinder. The cylindrical approxima-
corresponding dynamic response have been collected in the flight
tion of the airship hull can be justified by the fact that the highest
tests discussed in Section 3. Turbulent wind can be simulated in a
bending moment and shear force occur in the central part of the
wind tunnel and this technique has been used to measure the
hull [89,90]. Analysis accounting for the variation of cross-sectional
airship aerodynamics in turbulence. But even such data are limited
radius can be found in [89,90], where it is also shown that the
in literature.
difference in stresses obtained with and without the constant-
Lagrange [92] carried out an experimental study of the aero-
radius assumption are small ( o5%) near the middle of the hull.
dynamic force on a 0.76-m (30-in) scaled airship model in vertical
To use Eqs. (18) and (19), the bending moment and shear force
turbulence. Isotropic random turbulent flow was created in a wind
values must be first obtained. An empirical prediction method for
tunnel by using flow through a grid. The turbulence angle of attack
the maximum bending moment on the airship hull is given in [3]
was defined as ag ¼ wg =V, in which wg is the vertical gust velocity
which is useful for a preliminary check of internal pressure. A more
and V is the air speed. The following power spectral densities were
accurate calculation of bending moment and shear force requires
measured: of the turbulence input Fag ag , of the normal force
the external normal force distribution along the hull; it in turn can
coefficient FCN CN , and of the pitch moment coefficient FCM CM . The
be computed with the methods discussed in Section 2.
resulting transfer functions FCN CN =Fag ag and FCM CM =Fag ag for bare
hull and hull–fin combination are plotted in Fig. 16 against the
4.5. Summary wave number O ¼ 2pf =V, where f is the frequency in Hz. For the
bare hull, there is no peak in the pitch moment response but a peak
1
Based on the existing research, there is no strong evidence that for the normal force occurs at O ¼ 3:9 m1 (1:2 ft ). This wave
the flexibility effects have a significant impact on the flight number corresponds to a turbulence wave length of about twice the
characteristics of conventional modern airships. However, these hull length of the airship model. The addition of fins dramatically
effects could become important when the inflated hull structure changes the aerodynamic response to turbulence. The peak present
wrinkles or when the natural frequencies of the flexible modes in the normal force response of the bare hull disappears while there
become sufficiently low so as to interact with the rigid-body is a peak in the moment response curve. One can also observe an
modes, for example, if the airship were constructed of very thin increase in the steady value of CN (at O ¼ 0) and a decrease in the
materials. Furthermore, for unconventional airships with large steady value of CM due to the fins.
deformation, such as the Sanswire Stratellite airship in Fig. 10(a),
the aeroelastic effects must be taken into account in the dynamics 5.2. Theoretical methods
modeling.
Current analytical models for the aeroelastic analysis of airships are Two approaches have been used to incorporate atmospheric
limited and there are no effective tools to predict the aeroelastic effects turbulence into airship dynamics models. The first is an energy-
on the flight characteristics. Application of FEA and CFD codes is based method, where the turbulent air flow is assumed to be
hindered by their high computational cost. It is difficult to draw general inviscid and the force and moment due to the flow motion is
conclusions based on the above FEA and CFD works. At this point, many derived from the kinetic energy of the fluid. This method focuses on
outstanding questions remain, for example: What effect does structural the coupling between the flow motion and the added-mass
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 233

Fig. 16. Turbulence response of normal force and pitch moment in [92].

aerodynamic terms. Thus, it can be considered as an extension of the velocity at the location of the body-frame origin, i.e.,
the derivation of the added-mass force and moment in Section 2.2 2 3
@uf @vf @wf
where the added-mass terms are derived from kinetic energy via 6 @x @x
6 @x 7
7
Kirchhoff’s [30,17]. The second method is based on the slender 6 @uf @vf @wf 7
6 7
vf ,r ¼ vf þ UTv r, where Uv ¼ 6
@y 7
body theory. In this method, the turbulence is first incorporated ð22Þ
6 @y @y 7
into the aerodynamic computation for each cross-section along the 6 7
4 @uf @vf @wf 5
airship, and then the total force and moment are obtained by
@z @z @z
integrating the local force distribution over the airship. This
method can be considered as an extension of Munk’s equation where vf ¼ ½uf ,vf ,wf T is the flow velocity at the location of the body-
(1) or Jones and DeLaurier’s slice method in [27]. frame origin, r is the position vector of a point of interest from that
location, and Uv contains the spatial gradients of the non-uniform flow
velocity at the origin. All vectors in Eq. (22) are expressed in the body
5.2.1. Energy method
frame. Thomasson derived the change of the kinetic energy of the fluid
The energy-based method was proposed to investigate the
due to the presence of the vehicle and developed the motion equations
influence of flow turbulence on airship aerodynamics in the golden
using Lagrange’s equations. The results of [35] are summarized now,
age of airships. Taylor [93] derived the forces on a small stationary
but with some changes in form and notation for consistency of
body immersed in a non-uniform potential fluid from the fluid
presentation. First, the body frame in [35] can be located at an arbitrary
kinetic energy. A practical application of this work is the correction
position on the vehicle, while in this paper, the body frame is
of the axial drag measurement on old airship models in wind-
established at the CV and hence, the terms related to the position of
tunnel tests. It was found that the airship models tested in a wind
CV in [35] are dropped. Second, the fluid velocity in [35] is split into two
tunnel experience a horizontal force due to the spatial gradients of
terms, one representing a uniform current with no spatial gradients and
flow velocity as follows:
the other one representing a circulating flow with spatial gradients. In
@uf this paper, the two terms are combined in vf for a more compact
Ffx ¼ ð1 þ k1 Þmuuf ð20Þ
@x representation. Third, the fluid dynamic terms due to flow motion in
where uf is the flow velocity at the CV in the absence of the body. [35] are mixed with the inertial terms in the motion equations. Since
Recall that mu is the mass of the displaced air and k1 is the the inertial terms have already been presented in Eq. (10), they are not
translational added-mass factor along the x axis. It can be seen repeated in the following.
from Eq. (20) that the added-mass terms are coupled with the With the above changes, the force due to the fluid motion
spatial gradient of flow velocity. extracted from [35] is given as follows:
The fluid dynamic force on a body in a non-uniform potential Ff ¼ M11 v_ f þ x M11 vf þmuv_ f þ mux vf Uv M11 ðvvf Þ
fluid has also been studied in the field of hydrodynamics. Newman muUv ðvvf ÞUv M12 x ð23Þ
[31] extended Eq. (20) to account for the acceleration of an
unsteady fluid and the body velocity in the x direction u, yielding
  Recall that M11 and M12 are block matrices in the added-mass
@uf
Ffx ¼ ð1 þ k1 Þmu u_ f þðuf uÞ ð21Þ matrix in Eq. (2). It can be verified that the x component of the force
@x
in Eq. (23) reduces to the force in Eq. (21) for an axisymmetric
Lewis et al. [94] presented the equations of motion for an under- vehicle in pure translation, and it can be further reduced to Eq. (20)
water vehicle, incorporating the velocity and acceleration of the if the vehicle is stationary and the fluid flow is steady. The total
water flow. Some errors in [94] were found and corrected by potential fluid force is the sum of Ff in Eq. (23) and FA in Eq. (2),
Thomasson [35] in 2000 when he tried to apply the equations in respectively, i.e.,
[94] to airships. FA þ Ff ¼ M11 ðv _ x M11 ðvvf Þx M12 x
_ v_ f ÞM12 x
Thomasson [35] derived a new set of motion equations for a 
þ muv_ f þ mux vf Uv M11 ðvvf ÞmuUv ðvvf ÞUv M12 x
6-DOF rigid-body vehicle in an unsteady and non-uniform heavy
ð24Þ
fluid. In this work, the vehicle is assumed to be small so that the
changes in the velocity of the undisturbed flow over the length of
the vehicle are small relative to the flow velocity [35]. Under this In other words, to incorporate the force due to the flow motion, one
assumption, the flow velocity distribution in the absence of the can replace v_ with v
_ v_ f and v with vvf when computing FA and
vehicle can be obtained from the first-order Taylor expansion about then add the terms related to mu and Uv (the second line of Eq. (24)).
234 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

The total potential fluid moment is accordingly from [35]

V
_ ðv v
_ v_ f ÞM22 x
MA þ Mf ¼ M21 ðv f ÞM11 ðvvf Þ
o
ðv v  
f ÞM12 xx M21 ðvvf Þx M22 x ð25Þ x
z

Recall that M21 and M22 are also block matrices in the added-mass wg
matrix (see Eq. (2)). We can observe from Eq. (25) that the velocity
ww
gradient matrix Uv has no influence on the moment if the body
frame is established at the CV. As well, to incorporate the moment
due to the flow motion, one can simply replace v_ with v _ v_ f and v
with vvf in the calculation of MA . Fig. 17. Airship in vertical turbulence.
Azinheira et al. [95] investigated the wind influence on airship
dynamics via a Newtonian approach. Recently, they found that an space-dependent gust speed as
inertial moment term is missing in [95] and they corrected the
moment result via a Lagrangian approach [96]. In these two vf ,r ¼ vw þ vg ðrÞ ð27Þ
articles, Azinheira et al. incorporated the linear velocity and
Compared to Eq. (22), the second term in Eq. (27) can be an
acceleration (vf and v_ f ) and angular velocity and acceleration
arbitrary small disturbance. The mean wind speed vw is taken into
(xf and x _ f ) of the wind into the equations of motion. The angular
account by using the relative speed vvw to calculate the angle of
velocity xf represents the flow vorticity and can be obtained from
attack and dynamic pressure in the aerodynamic computation, for
the gradient matrix Uv . The matrix Uv is split into symmetric and
example, in Eqs. (1), (5), or (7). The gust vg is incorporated as an
antisymmetric parts, with the symmetric representing the irrota-
additional term, as discussed below.
tional strain rates and the antisymmetric representing the vorti-
Calligeros et al. [97] and Lagrange [92] calculated the normal
city. Readers are referred to literature on atmospheric turbulence
force for an airship traveling with a forward speed V through a
modeling, such as [91], for how to obtain xf from Uv . The force
vertical gust represented by wg (see Fig. 17). The normal force on a
result in [95] is in agreement with Thomasson’s equation (23),
local cross-section due to the vertical gust is [92]
except that the matrix M12 is assumed to be zero and the force
related to the spatial gradients of wind velocity is not considered in dF Ng dS
¼ 2q0 ag ð28Þ
[95]. However, additional terms due to xf and x _ f appear in the de de
moment result presented in [96]
where recall that ag ¼ wg =V. Eq. (28) is similar to Eq. (1), except that a
Mf ð x f , x _ f þ x ðJu þ M22 Þxf
_ f Þ ¼ ðJu þ M22 Þx ð26Þ is replaced by a small gust-induced angle of attack ag and that the
where Ju is the inertia tensor of the displaced air. Therefore, besides added-mass factors are not incorporated. Lagrange [92] compared the
the terms in Eq. (25), one can add the terms in Eq. (26) to account for frequency responses of normal force and pitch moment calculated
the vorticity of the wind. Azinheira et al. analyzed the influence of from Eq. (28) to wind-tunnel test results in Fig. 16, but considerable
constant wind in the longitudinal and lateral directions on the error was observed. This was likely because the airship was not
stability modes of an airship moving with low speeds [95]. Under slender (with fineness ratio of 4) and the added-mass factors were not
longitudinal wind, the frequency of pitch oscillation mode included to account for the finite length in Eq. (28).
decreases and the frequency of the roll oscillation mode increases DeLaurier et al. [98] and Layton et al. [99] extended the
as the longitudinal wind speed increases. Under lateral wind, the aerodynamic computational approach in [27] to incorporate the
lateral modes are insensitive to the wind, but the pitch oscillation gust into the aerodynamics model. The normal force distribution
exhibits a lower damping ratio and a lower frequency as the wind due to the gust is obtained as [98]
speed increases. However, we caution readers to be careful in using
dF Ng dS h   a a

g ai
these stability results since the inertial moment term in [95] is ¼ ðk2 k1 ÞZk q0 sin 2a þ 2ag cos þ sinð2aÞcos
de de 2 2 2
incorrect.
ð29Þ
The following limitations of the energy-based method
employed in [35,95,96] should be noted. First, the aerodynamics Comparing to Eq. (7), we can see that the influence of small gust on the
due to viscosity and the lift and drag on tail fins are not considered viscous term is ignored here. If a ¼ 0, ag is small, and the factor
in these works. To account for the flow motion, relative velocity ðk2 k1 ÞZk is assumed to be unity, Eq. (29) reduces to Eq. (28).
between the body and the air should be used in the calculation of DeLaurier et al. [98] calculated the frequency responses of aero-
these aerodynamic terms. Second, Eqs. (23)– (26) are derived based dynamic force, velocities, bending moment, and shear force to
on a small-body assumption. The applicability of these equations random gust for the Akron airship. It was found that the peak normal
to large airships is uncertain. Third, Eqs. (23)–(26) are ‘‘point’’ equa- force and pitch moment due to gust occur at wave number
tions and do not provide the force distribution along an airship in O ¼ 0:033 m1 (0:01 ft1 ) and 0:016 m1 (0:005 ft1 ), respectively.
turbulent wind, which is required for structural analysis. These wave numbers correspond to wave lengths of the order of
magnitude of the airship length (239 m). The motion response is
5.2.2. Slender body method dependent upon the control gains for the elevator deflection, implying
Based on the slender body theory, the force on a cross-section of that atmospheric turbulence should be considered in the controller
a slender body can be calculated from the cross flow in the plane of design. In addition, the simulation results in [98] show that the gust
that cross-section. Then, the aerodynamic force on a cross-section has only a small influence on the bending moment and shear force.
of an airship moving in a turbulent wind can be obtained from the Compared with the energy-based method, the slender body-
local turbulent flow in that cross-sectional plane. Compared to the based method has an advantage that the turbulence field can be
inviscid flow assumption of the energy-based approach, the slender described more accurately relative to Eq. (22). Furthermore,
body theory allows for a more general modeling of the turbulent because the turbulence is incorporated through the normal force
wind field. distribution, the slender body-based method can be used to obtain
As illustrated in Fig. 17, the turbulent wind can be represe- the shear force and bending moment distributions resulting from
nted as a superposition of a mean wind speed and a small atmospheric turbulence.
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 235

On the other hand, the modeling employed in [97–99] implies a slice due to the turbulence can be obtained using the energy-based
frozen-field turbulence, i.e., the turbulent velocities do not change method. Then, the total force and moment could be computed by
during the time of passage of the aircraft. This is true for most flight summation of those on the slices.
conditions of airplanes [91] but has been argued inappropriate for
LTA aircraft due to their much lower flight speeds [63]. As such, the
flow acceleration (v_ w and v_ g ) may need to be considered in airship 6. Incorporation of effects of ballonets
dynamics.
Ballonets are components particular to LTA aircraft, used to
5.3. Summary control buoyancy and to maintain sufficient pressure difference
between the surrounding air and the internal lifting gas across the
The articles [35,95,96] have advanced our understanding of the hull envelope. Typically two or three in number, they are air bags
coupling between the flow motion and the added-mass aerodynamic contained inside the hull, inflated or deflated by means of a blow
terms. The energy-based method employed in these papers charac- fan as drawn in Fig. 18. Pressure height is the airship height at
terizes an inviscid fluid by the flow velocity, acceleration, and spatial which all the ballonets are empty and the internal pressure can no
velocity gradients at a single point (the CV of an airship which is small longer be controlled by the deflation or inflation of ballonets [3]. To
compared to the turbulence wave length). In the slender body-based reach a pressure height of 4 km, the ballonet can occupy up to 50%
method used in [97–99], the flow field is decomposed into mean wind of the hull volume at low altitude [100]; for a stratospheric airship
velocity and a space-dependent gust velocity, and then the force on designed to fly at an altitude of approximately 20 km, the ballonet
each cross-section of the airship is computed from the gust angle of volume can be 90% of the hull volume at sea level [101]. These
attack at that position. This method can be used for large airships but numbers imply that the ballonets could strongly affect the
the flow acceleration should be incorporated. A possible scenario to dynamics characteristics of an airship, particularly if they move
extend the energy-based approach to large airships would be to or deform as a result of airship motion. The influence of ballonets on
discretize the airship into a series of small slices and introduce a airship dynamics depends on their size and shape and how they are
multi-point turbulence model to describe the turbulence velocity and attached to the hull envelope. In addition, the airship becomes a
acceleration at the center of each slice. The force and moment on each variable mass system due to the inflation and deflation of ballonets.
To date, a very limited number of experimental and theoretical
studies on ballonets can be found in literature.

6.1. Experimental studies

Nakadate [102] conducted ballonet tilting tests for the SPF-2

airship with a water model and an air/helium model, with the
ballonet filled in stages up to 100% fullness. It was found that the
shape of the ballonet is preserved without significant deformation
at different pitch angles, and matches reasonably well with FEA
results from MSC.Dytran. The variation of ballonet volume and the
hull internal pressure at different altitudes have also been mea-
Fig. 18. Pressure control system in a non-rigid airship [3]. sured in a flight test of the SPF-2 [102], as displayed in Fig. 19(a).

Fig. 19. (a) Altitude, ballonet volume, and hull pressure for the SPF-2 airship [102] and (b) ballonet CV position for the SN-200 aerostat [103].
236 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

The ballonets are filled to approximately 80% of their full size at an ejection from the ballonets, i.e., [68]
altitude of 1000 m and reduced to 20% at 4000 m, so that the @r _
differential pressure between the lifting gas and the surrounding _ b V ¼ VB
Fb ¼ m HV ð31Þ
@H
air is maintained within the structural limits. Another observation
is that the ballonets did not experience a sloshing problem during where mb is the mass of the ballonet air. Because the altitude
the flight tests [102,100]. changes slowly during the ascent (with an increase of about 20 km
DeLaurier [103] reported the experimental data of the ballonet in 5 h) in [68], the force Fb is small and has minor influence on the
CV position for the 50-m (163-ft) SN-204 aerostat, at different pitch airship dynamics.
angles (see Fig. 19(b)). Tests were carried out for three cases with Note that the relative motion between the ballonets and the hull is
air occupying 1/3, 2/3, and the whole of the ballonet envelope. It not considered in [104,68]. In the work of Maekawa and Saito [101],
can be seen from Fig. 19(b) that the ballonet CV location is affected the ballonet is assumed to be a cylindrical container and the internal
by the aerostat pitch angle. In particular, with the ballonet 1/3 full, air to be the sloshing fluid in it. With this assumption, the motion of
the variation of the ballonet CV location is as high as 20% of the the ballonet sloshing can be analyzed using the dynamics models that
aerostat length. This indicates the coupling between ballonet have been developed for the sloshing liquids in moving containers
motion and pitch rotation. These data were used for the parameter [105,106]. As illustrated in Fig. 20(a), the fluid is modeled in two parts:
identification of a semi-empirical model for ballonet sloshing one sloshing in the container (represented by an infinite number of
[103], which will be further discussed in Section 6.3. oscillating masses m1 ,m2 , . . . ,mn ) and one fixed relative to container
(represented by a mass mb0 and a moment of inertia Ib0 ). The
analytical expressions for m1 ,m2 , . . . ,mn can be found in [105,106].
6.2. Theoretical models It is found that m2 ,m3 , . . . ,mn are much smaller than m1 , and thus, can
be neglected [101]. To incorporate the ballonet sloshing into the
Cai et al. [104] formulated the equations of motion for an airship longitudinal linear dynamics model, a new differential equation is
with the ballonets modeled as points at fixed positions and variable written to describe the motion of m1 as [101]
masses. The variation of the mass matrix (see Eq. (10)) due to the y€ 1 þ 2z1 o1 y_ 1 þ o21 y1 ¼ u_ þ ðh1 lzb Þqg
_ y ð32Þ
ballonet inflation and deflation is taken into account in their
equations. That is, the total mass, position vector of CG, and the where y1 is the displacement of m1 from its equilibrium position, z1 is
second moment of inertia are written as [104] the damping ratio, o1 is the undamped natural frequency, h1 is the
vertical position of m1 from the ballonet CG (see Fig. 20(a)), and lzb is
Nb
X Nb Nb
1 X X vertical distance between the ballonet CG and the hull CV. The
m¼mþ Dmbk , rG ¼ r G þ Dmbk rbk , J ¼ J Dmbk r 
bk rbk
k¼1
mk¼1 k¼1
analytical expressions for o1 and h1 can be found in [105,106]. The
right hand side of Eq. (32) represents the excitation of the sloshing
ð30Þ
mass due to the airship motion. Then, the reaction force (in the x
where Nb is the number of ballonets, rbk are the positions of the direction) and moment (in the y direction) from the ballonet to the
ballonets, Dmbk are the change of the ballonet masses due to airship are incorporated into the linear dynamics model of the airship
inflation or deflation, m, r G , and J are the initial mass, position of CG, (the first equation in (11)).
and inertia tensor of the airship before the ballonets are inflated or The simulated responses of u_ and y€1 to elevator step input (see
deflated. No simulation results are presented in [104] to demon- Fig. 20(b)) are used to evaluate the influence of ballonet sloshing on
strate the ballonet influence on airship dynamics. the rigid-body motion of a 25 m airship [101] with two ballonets. In
Mueller et al. [68] incorporated the mass variation of ballonets that study, three cases for the ballonet dimensions are considered.
into the point-mass dynamics model for the ascent trajectory In the first (nominal) case, each ballonet occupies about 6.5% of the
optimization of stratospheric airships. Under neutral buoyancy envelope volume, the sloshing mass m1 of each ballonet is about
condition, the total mass of the airship including the ballonets 18% of the ballonet mass mb , the damping ratio z is 0.01, and the
depends on the hull volume as rVB , where the air density r is a sloshing frequency o1 is 0.53 Hz, which is approximately five times
function of the altitude H. Recall that VB is the total airship volume. the airship’s longitudinal pendulum frequency (see Section 3.3) of
Then, Mueller et al. incorporated an additional force due to the air 0.1 Hz. The ballonet slosh has only a minor influence on the airship

Fig. 20. (a) Ballonet sloshing model and (b) simulation results in [101].
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 237

rigid-body motion. If the ballonet shape becomes squatter (with considered the least stable and most important mode for design
the diameter doubled and the height reduced to 1/4 of the nominal purposes.
case), the sloshing mass becomes bigger (about 76% of mb ) and the The modeling method in [101,103] are different in two respects.
sloshing frequency is reduced to 0.27 Hz. The sloshing vibration First, the sloshing motion of the lifting gas is considered in [103] but
exhibits a larger amplitude than that of the nominal case and there not in [101]. Second, the estimation methods for the sloshing
is a more noticeable effect on the rigid-body motion. If the ballonet damping and stiffness terms are different. In [101], the sloshing
is of the nominal height and twice the diameter (third case), frequency o1 is obtained from the analytical solution for sloshing
it occupies 53% of the envelope volume. In this case, the sloshing liquids in containers, and the damping ratio z1 is arbitrarily chosen
frequency is 0.37 Hz and m1 is about 36% of mb , and it is observed in the range 0.01–0.1. In [103], the equivalent stiffness K^ and
that u_ oscillates with the largest amplitude due to the ballonet damping coefficient C^ are obtained by employing parameter
sloshing. The simulation study in [101] demonstrates that the identification on experimental data.
influence of ballonet sloshing on the airship motion depends on the Different sloshing characteristics of ballonets are obtained in
geometry of the ballonet. [103,101]. First, the equivalent sloshing mass M ^ is larger than the
ballonet mass mb in [103], while the sloshing mass m1 is only 18% of
6.3. Semi-empirical models the ballonet mass in the nominal case in [101]. Second, the
undamped natural frequency of the ballonet sloshing in [101] is
A semi-empirical method has been applied to investigate the as high as five times of the airship’s longitudinal pendulum
influence of ballonet motions on the longitudinal stability of frequency, while the natural frequency of the ballonet sloshing
tethered aerostats [103]. This method accounts for the slosh of is only 1.4 times of aerostat’s longitudinal pitch mode frequency in
ballonet air and lifting gas, both of which are modeled as lumped [103]. Finally, the ballonet influence on the aerostat motion is
masses (mb and mg ) acted on by a spring and a dashpot, as shown in found to be considerable in [103], while this influence is rather
Fig. 21. To incorporate the ballonet motions into the linear small for the airship in [101].
dynamics model of the tethered aerostat, a motion equation is
derived for mb as [103] 6.4. Summary
   
M^ x€ b þ C^ x_ b þ K^ xb ¼  mb mg Vb u_
V
mb lzb mg lzg b q_ To understand the ballonet influence on airship dynamics, two
Vg Vg
  relevant issues have been considered: the variation of mass due to
Vb inflation and deflation of ballonets, and the motion of ballonets
 mb mg gy ð33Þ
Vg relative to the hull.
For flight in a large range of altitudes, the volume of ballonet air
where xb denotes the displacement of mb from its equilibrium changes dramatically. Thus, the effects of ballonet mass variation
position, Vb is the volume of ballonet air and Vg is the volume of on the total mass matrix of the airship must be considered in the
lifting gas, lzb and lzg are the vertical distances from the hull CV to dynamics modeling. On the other hand, if the airship ascent speed
the CG of the ballonet air and lifting gas, M, ^ C^ , and K^ are an is low, the force due to the air ejection from the ballonets can be
equivalent mass, damping coefficient, and stiffness, respectively. ignored.
The mass term is M ^ ¼ mb þ mg ðVb =Vg Þ2 and the stiffness K^ and The modeling of the ballonet slosh relative to the hull is a more
damping coefficients C^ are determined from experiments, as complicated problem. Spring-mass-dashpot models have been
mentioned in Section 6.1. Then, the reaction forces and moment proposed for the ballonet slosh modeling and the induced reaction
from the ballonet and the gas are added to the corresponding force forces transmitted to the airship have been studied. One difficulty
and moment terms into the linear dynamics model of the aerostat. lies in determining the parameters of these spring-mass-dashpot
Simulation results for the SN-204 aerostat in [103] show that models. Some analytical results available for sloshing liquids in
the ballonet motion has a strong impact on the modes of the rigid containers can be used to compute the sloshing frequency for
tethered aerostat, in particular, the pitching of the aerostat couples a preliminary check against the longitudinal pendulum frequency
with the ballonet slosh. The first mode of the tethered aerostat is a of the airship. However, their utility is limited, since the ballonet
tether-induced pendulum mode and is unchanged by the ballonet. slosh frequency is sensitive to the shape of the ballonet envelope
The second mode is related to the pitch rotation and while the and to how the ballonet and hull envelopes are attached together.
damping of this mode is unchanged by the ballonet, its frequency A semi-empirical method where the model parameters are identi-
decreases to half of its original value due to the ballonet sloshing. fied from experimental data, as in [103], is likely to produce a more
Furthermore, a new third mode is introduced by the ballonet. accurate model of ballonet sloshing. Generally speaking, the
Although this new mode is stable, its damping ratio is of the same modeling of ballonet sloshing is still not well understood and
order of magnitude as of the first aerostat mode, which has been requires further research supported by experimentation.

7. Conclusions

A state-of-the-art literature review on airship dynamics mod-

eling is presented. The references are categorized according to the
major topics in this area: aerodynamics, flight dynamics, incor-
poration of structural flexibility, incorporation of atmospheric
turbulence, and effects of ballonets. Relevant analytical, numerical,
and semi-empirical techniques are discussed, pointing out differ-
ences between LTA and HTA aircraft.
It should be noted that the five topics addressed in this review
are interrelated, as illustrated by the diagram in Fig. 22. Collar
suggested to use a triangle of inertial, elastic, and aerodynamic
Fig. 21. DeLaurier’s model for ballonet sloshing [103]. forces to visualize disciplines of flight dynamics, structural
238 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239

Inertial forces [6] Dorrington GE. Development of an airship for tropical rain forest canopy
exploration. Aeronautical Journal 2005;109:361–72.
[7] Hygounenc E, Jung I, Soueres P, Lacroix S. The autonomous blimp project of
Atmospheric LAAS-CNRS: achievements in flight control and terrain mapping. The Inter-
turbulence Ballonet national Journal of Robotics Research 2004;23(4–5):473–511.
[8] Kulczycki EA, Joshi SS, Hess RA, Elfes A. Towards controller design for
autonomous airship using SLC and LQR Methods. In: AIAA Guidance,
navigation and control conference and exhibit, Keystone, Colorado, August
21–24, 2006.
[9] Elfes A, Bueno SS, Bergerman M, Paiva ECD, Ramos JG, Azinheira JR. Robotic
airships for exploration of planetary bodies with an atmosphere: autonomy
Aerodynamic forces Elastic forces challenges. Autonomous Robots 2003;14(2–3):147–64.
[10] Kusagaya T, Kojima H, Fujii HA. Estimation of flyable regions for planetary
Fig. 22. The interrelation of the five topics. airships. Journal of Aircraft 2006;43(4):1177–81.
[11] Pope C. The big lift. Professional Engineering 2004;17(8):24–5.
[12] Schmidt DK. Modeling and near-space stationkeeping control of a large high-
vibrations, and aeroelasticity [76]. For example, in flight dynamics, altitude airship. Journal of Guidance, Control and Dynamics 2007;30(2):
one studies the interaction between inertial and aerodynamic 540–7.
[13] Lee S, Bang H. Three-dimensional ascent trajectory optimization for strato-
forces, and aeroelasticity deals with the interaction between spheric airship platforms in the jet stream. Journal of Guidance, Control and
aerodynamic and elastic forces. We have included atmospheric Dynamics 2007;30(5):1341–52.
turbulence and ballonet effects into the diagram because of their [14] Lee Y-G, Kim D-M, Yeom C-H. Development of Korean high altitude platform
systems. International Journal of Wireless Information Networks 2006;13(1):
importance for airship dynamics. The thicknesses of the lines and
31–42.
circles quantify the amount of existing research in each discipline [15] Etkin B. Dynamics of flight: stability & control. 3rd ed.. New York: Wiley;
for airships. Many airship dynamics analyses, either experimental 1996. p. 93–104, 156–9.
or theoretical, are focused on aerodynamics and flight dynamics. [16] Liao L, Pasternak I. A review of airship structural research and development.
Progress in Aerospace Sciences 2009;45(4–5):83–96.
Although the construction of modern airships using inflated [17] Fossen TI. Guidance and control of ocean vehicles. New York: Wiley; 1998
membrane structures is a major difference from HTA aircraft, p. 5–56.
research on the coupling between elastic forces and aerodynamic [18] Jones R. The aerodynamical characteristics of the airship as deduced from
experiments on models, with application to motion in a horizontal plane.
and inertial forces is limited. The modeling of atmospheric Journal of Royal Aeronautical Society 1924;28(158):88–150.
turbulence has been studied in detail for the flight dynamics of [19] Curtiss HC, Hazen DC, Putman WF. LTA aerodynamic data revisited. Journal of
HTA aircraft. With researchers’ recent effort, the methodology to Aircraft 1976;13(11):835–44.
[20] H.B. Freeman, Force measurements on a 1/40-scale model of the U.S. Airship
incorporate the turbulent wind into airship aerodynamic and flight Akron. NACA TR-432; 1933.
dynamic models has been improved. Since atmospheric turbulence [21] Freeman HB. Pressure distribution measurements on the hull and fins of a
can produce a high frequency disturbance on the system, it may 1/40-scale model of the U.S. Airship Akron. NACA TR-443; 1934.
[22] Zahm AF. Air forces, moments and damping on model of fleet airship
excite structural vibrations which in turn may influence the
Shenandoah. NACA TR-215; 1926.
airship’s dynamics, and this has not been investigated in the [23] Jones R, Bell AH. Experiments on a model of the airship R.101. ARC RM-1168;
literature. Researchers have demonstrated that the ballonets’ 1926.
motions induce reaction forces and moments on the airship. The [24] Jones R, Bell AH. The distribution of pressure over the hull and fins of a model
of the rigid airship R.101, and a determination of the hinge moments on the
ballonets’ motion relative to the hull is still difficult to model control surfaces. ARC RM-1169; 1926.
accurately. If the internal lifting gas pressure is high enough to [25] Gomes SBB. An investigation into the flight dynamics of airships with
maintain the hull shape during the flight, it is reasonable to assume application to the YEZ-2A. PhD thesis, Cranfield Institute of Technology;
1990.
that the ballonets have no influence on the aerodynamics. Thus, [26] Curtiss HC, Hazen DC, Putman WF. Experimental investigations on hull–fin
there are no lines connecting aerodynamic force and ballonets in interferences of the LOTTE airship. Aerospace Science and Technology
the diagram in Fig. 22. Existing studies on ballonet motion show 2003;7(8):603–10.
[27] Jones SP, DeLaurier JD. Aerodynamic estimation techniques for aerostats and
that the ballonet sloshing frequency is higher than that of the airships. Journal of Aircraft 1983;20(2):120–6.
airship rigid-body modes and may approach the lowest structural [28] McCormick BW. Aerodynamics, aeronautics and flight mechanics. 2nd ed..
vibration frequency, causing an interplay between the two New York: Wiley; 1995. p. 99–116, 160–5.
[29] Munk MM. The aerodynamic forces on airship hulls. NACA TR-184; 1924.
motions. However, no research has been published to date on
[30] Lamb H. Hydrodynamics. 6th ed.. New York: Dover; 1945. p. 160–201.
the interaction of ballonet motion and structural deformations. [31] Newman JN. Marine hydrodynamics. Cambridge, MA: MIT Press; 1977
This literature review may also be useful for the analysis of p. 132–49.
unconventional airships, most of which are still only in the [32] Lutz T, Fund P, Jakobi A, Wagner S. Summary of aerodynamic studies on the
lotte airship. In: Proceeding of the 4th international airship convention and
conceptual design stage. Other than the hybrid heavy lift airship exhibition, Cambridge, UK, July 28–31, 2002.
described in Section 3, few works have been published on the [33] Blevins RD. Formulas for natural frequency and mode shape. New York:
dynamics of unconventional airships. Although existing airship Robert E. Krieger Publishing Company; 1979.
[34] Li Y, Nahon M. Modeling and simulation of airship dynamics. Journal of
dynamics modeling techniques reviewed in this manuscript may Guidance, Control and Dynamics 2007;30(6):1691–700.
be inadequate for unconventional airships, the review should [35] Thomasson PG. Equations of motion of a vehicle in a moving fluid. Journal of
provide a starting point for further advancement of these methods. Aircraft 2000;37(4):630–9.
[36] Allen HJ, Perkins EW. Characteristics of flow over inclined bodies of
revolution. NACA RM-A50L07; 1951.
[37] Hopkins EJ. A semi-empirical method for calculating the pitching moment of
References bodies of revolution at low mach numbers. NACA RM-A51C14; 1951.
[38] Hoak DE. USAF stability and control DATCOM. Flight Control Division, Air
Force Dynamics Laboratory, Wright-Patterson Air Force Base, OH; 1965
[1] Gerken LC. Airships: history and technology. Chula Vista, CA: American [Section 4.2].
Scientific Corporation; 1990. p. 1–19. [39] Jorgensen LH. Prediction of static aerodynamic characteristics for space-shuttle-
[2] Wilson JR. A New era for airships. Aerospace America 2004;42(5):27–31. like and other bodies at angles of attack 01 to 1801. NASA TN-D-6996; 1973.
[3] Khoury G, Gillett J. Airship technology. Cambridge, UK: Cambridge University [40] Peddiraju P, Liesk T, Nahon M. Dynamics modeling for an unmanned,
Press; 1999. p. 4–17, 28–39, 141–209, 475–504. unstable, fin-less airship. In: 18th AIAA lighter-than-air technology con-
[4] Oh S, Kang S, Lee K, Ahn S, Kim E. Flying display: autonomous blimp with real- ference, Seattle, Washington, May 4–7, 2009.
time visual tracking and image projection. In: Proceedings of the IEEE/RSJ [41] Pitts WC, Nielsen JN, Kaattari GE. Lift and center of pressure of wing-body-tail
international conference on intelligent robots and systems, Beijing, China, combinations at subsonic, transonic and supersonic speeds. NACA TR-1307;
2006. p. 131–6. 1957.
[5] Colozza A, Dolce JL. High-altitude, long-endurance airships for coastal [42] DeLaurier JD. A stability analysis for tethered aerodynamically shaped
surveillance. NASA TM-2005-213427; 2005. balloons. Journal of Aircraft 1972;9(9):646–51.
 Y. Li et al. / Progress in Aerospace Sciences 47 (2011) 217–239 239

[43] Lambert C, Nahon M. Stability analysis of a tethered aerostat. Journal of [75] Hunt JD. Structural analysis of aerostat flexible structure by the finite element
Aircraft 2003;40(4):705–15. method. Journal of Aircraft 1982;19(9):674–8.
[44] Hoerner SF. Fluid-dynamic drag. NJ: Midland Park; 1958. p. 1–19. [76] Dowell EHD. A modern course in aeroelasticity. 4th ed.. Dordrecht, Nether-
[45] Dorrington GE. Drag of spheroid-cone shaped airship. Journal of Aircraft lands: Kluwer Academic Publishers; 2005.
2006;43(2):363–71. [77] Meirovitch L, Tuzcu I. Unified theory for the dynamics and control of
[46] Jakobi A, Lutz T, Wagner S. Calculation of aerodynamic forces on inclined maneuvering flexible aircraft. AIAA Journal 2004;42(4):714–27.
airship bodies—boundary layer calculation method. In: 3rd international [78] Sanswire-Tao-Products-StratelliteTM, Retrieved from /http://www.sanswir
airship convention and exhibition, Friedrichshafen, Germany, July 1–5, 2000. etao.com/product-stratellite.htmS on June 30; 2010.
[47] Lutz T, Funk P, Jakobi A, Wagner S. Aerodynamic investigations on inclined [79] Smith MS, Rainwater EL. Applications of scientific ballooning technology to
airship bodies. In: 2nd International airship convention and exhibition, high altitude airships. In: 3rd AIAA annual aviation technology, integration,
Bedford, UK, July 26–28, 1998. and operations technical forum, Denver, CO, November 17–19, 2003.
[48] Jakobi A, Funk P, Lutz T, Wagner S. Modelling of airship wakes applying higher [80] Bessert N, Frederich O. Nonlinear airship aeroelasticity. Journal of Fluids and
order panel elements. In: 14th AIAA lighter-than-air technical committee Structures 2005;21:731–42.
convention and exhibition, Akron, OH, July 15–19, 2001. [81] Burgess CP. Airship design. The Ronald Press Company; 1927. p. 109–52.
[49] Lutz T, Leinhos D, Jakobi A, Wagner S. Theoretical investigations of the [82] Amiryants GA, Grigorive VD, Ishmuratov FZ, Franz A, d’Henin E, Kaempf B.
flowfield of airships with a stern propeller. In: Proceedings international Investigations of airship aeroelasticity. In: 23rd international congress of
airship convention and exhibition, Bedford, UK, July 5–7, 1996. aerospace sciences, Toronto, Canada, 2002.
[50] Lutz T, Funk P, Jakobi A, Wagner S. Calculation of the propulsive efficiency for [83] Omari K, Schall E, Koobus B, Dervieux A. Inviscid flow calculation around a
airships with stern thruster. In: 14th AIAA lighter-than-air technical com- flexible airship. In: 8th conference of applied mathematics and statistics, Jaca,
mittee convention and exhibition, Akron, OH, July 15–19, 2001. Spain, September 15–18, 2004.
[51] Lutz T, Wagner S. Drag reduction and shape optimization of airship bodies. [84] Liu J-M, Lu C-J, Xue L-P. Investigation of airship aeroelasticity using fluid–
Journal of Aircraft 1998;35(3):345–51. structure interaction. Journal of Hydrodynamics 2008;20(2):164–71.
[52] Nejati V, Matsuuchi K. Aerodynamics design and genetic algorithms for [85] Venkatesan C, Friedmann P. Aeroelastic effects in multi-rotor vehicles with
optimization of airship bodies. JSME International Journal. Series B: Fluids application to a hybrid heavy lift system. Part I: formulation of equations of
and Thermal Engineering 2003;46(4):610–7. motion. NASA CR-3822; 1984.
[53] Wang X-L, Shan X-X. Shape optimization of stratosphere airship. Journal of [86] Bennaceur S, Azouz N, Boukraa D. An efficient modelling of flexible airships.
Aircraft 2006;43(1):283–7. In: 8th Biennial ASME conference on engineering systems design and
[54] Kale SM, Joshi P, Pant RS. A generic methodology for determination of drag analysis, Torino, Italy, July 4–7, 2006. p. 573–82.
coefficient of an aerostat envelope using CFD. In: 5th AIAA aviation, [87] Li Y, Nahon M, Sharf I. Modeling and simulation of flexible airships. AIAA
technology, integration, and operations conference, Arlington, VA, September Journal 2009;47(3):592–605.
26–28, 2005. [88] Vorus WS, Hylarides S. Hydrodynamic added-mass matrix of vibrating ship
[55] Wong KY, ZhiYung L, DeLaurier J. An application of source-panel and vortex based on a distribution of hull surface sources. Transactions of the Society of
methods for aerodynamic solutions of airship configurations. In: 6th AIAA Naval Architects and Marine Engineers 1981;89:397–416.
lighter-than-air technology conference, Norfolk, VA, June 26–28, 1985. [89] Kraska M. Structural analysis of the CL-160 airship. In: 14th AIAA lighter-
p. 78–83. than-air technical committee convention and exhibition, Akron, OH, July
[56] Pannell JR, Bell AH. Experiments on rigid airship R.29. ARC RM-675; 1920. 15–19, 2001.
[57] Thompson FL. Full-scale turning characteristics of the U.S.S. Los Angeles. [90] Chen W-J, Xiao W-W, Kroplin B, Kunze A. Structural performance evaluation
NACA TR-333; 1930. procedure for large flexible airship of HALE stratospheric platform concep-
[58] Bailey DB. Patrol Airship Concept Evaluation (PACE): Final Report, NADC- tion. Journal of Shanghai Jiaotong University 2007;E-12(2):293–300.
85019-60, 1985. [91] Etkin B. Turbulent wind and its effect on flight. UTIAS Review 44, University of
[59] Jex HR, Gelhausen P. Pre- and post-flight-test models versus measured Toronto; 1980.
skyship-500 control responses. In: 7th AIAA lighter-than-air technology [92] Lagrange MJB. Aerodynamic forces on an airship hull in atmospheric
conference, Monterey, CA, August 17–19, 1987. p. 87–97. turbulence. Technical Report 277, Institute for Aerospace Studies, University
[60] Kornienko A. System identification approach for determining flight dynami- of Toronto; 1984.
cal characteristics of an airship from flight data. PhD thesis, University of [93] Taylor GI. The forces on a body placed in a curved or converging stream of fluid.
Stuttgart; August 2006. Proceeding of the Royal Society of London, Series A 1928;120(785):260–83.
[61] Flax AH. Comment on ‘A comparison of different forms of dirigible equations [94] Lewis DJG, Lipscombe JM, Thomasson PG. The simulation of remotely
of motion’. Journal of Guidance, Dynamics, and Control 1979;2(6):544. operated underwater vehicles. In: ROV’84 conference and exposition, San
[62] Amann JH. A comparison of a nonlinear flight dynamic simulation of an Diego, CA, 1984. p. 245–51.
airship with flight test results. In: 7th AIAA lighter-than-air technology [95] Azinheira JR, de Paiva EC, Bueno SS. Influence of wind speed on airship
conference, Monterey, CA, August 17–19, 1987. p. 78–86. dynamics. Journal of Guidance, Control and Dynamics 2002;25(6):1116–24.
[63] Ringland RF, Tischler MB, Jex HR, Emmen RD, Ashkenas IL. Flight dynamics [96] Azinheira JR, de Paiva EC, Bueno SS. Erratum on ‘Influence of wind speed on
analysis and simulation of heavy lift airships. NACA CR-16647; 1982. airship dynamics’. Journal of Guidance, Control and Dynamics 2008;31(2):
[64] Cook MV, Lipscombe JM, Goineau F. Analysis of the stability modes of the 443–4.
non-rigid airship. The Aeronautical Journal 2000;104(1036):279–90. [97] Calligeros JM, McDavitt PW. Response and loads on airships due to discrete
[65] Yamasaki T, Goto N. Identification of blimp dynamics via flight tests. and random gusts. Technical Report 72-1, Aeroelastic and Structures
Transactions of the Japan Society for Aeronautical and Space Sciences Research Laboratory, MIT; 1958.
2003;46(153):195–205. [98] DeLaurier JD, Hui KCK. Airship survivability in atmospheric turbulence. In:
[66] Li Y. Dynamics Modeling and simulation of flexible airships. PhD thesis, AIAA lighter-than-air system technology conference, Annapolis, MD, 1981.
McGill University; 2008. p. 48–61.
[67] Bestaoui Y, Hima S. Some insights in path planning of small autonomous [99] Layton DM, Wrobleski JJ. The lateral response of an airship to turbulence. In:
blimps. Archives of Control Systems 2001;11(3–4):139–66. AIAA lighter-than-air system technology conference, Anaheim, CA, July
[68] Meuller JB, Zhao YJ, Garrard WL. Optimal ascent trajectories for stratospheric 25–27, 1983. p. 116–23.
airships using wind energy. Journal of Guidance, Control, and Dynamics [100] Maekawa S, Nakadate M, Takegaki A. Structures of the low-altitude stationary
2009;32(4):1232–45. flight test vehicle. Journal of Aircraft 2007;44(2):662–6.
[69] Miele A. Flight mechanics. Volume 1: theory of flight paths. Addison-Wesley; [101] Maekawa S, Saito K. The effect of ballonet slosh on an airship’s longitudinal
1962. p. 42–68. motion. Transactions of the Japan Society for Aeronautical and Space Sciences
[70] Slegers N, Brown AX. Comment on ‘Three-dimensional ascent trajectory 2004;47(155):44–50.
optimization for stratospheric airship platforms in the jet stream’. Journal of [102] Nakadate M. Development and flight test of SPF-2 low altitude stationary
Guidance, Control, and Dynamics 2009;32(5):1692–3. flight test vehicle. In: 5th AIAA aviation, technology, integration, and
[71] Evans JR, DeLaurier JD. The Shenandoah flies again: a computer simulation. operations conference, Arlington, VA, September 26–28, 2005.
In: AIAA lighter-than-air systems technology conference, Annapolis, MD, July [103] DeLaurier J. Influence of ballonet motions on the longitudinal stability of
8–10, 1981. p. 62–73. tethered aerostats. Journal of Aircraft 1980;17(5):305–12.
[72] Kang W, Suh Y, Woo K, Lee I. Mechanical property characterization of film- [104] Cai Z-L, Qu W-D, Xi Y-G. Dynamic modeling for airship equipped with
fabric laminate for stratospheric airship envelope. Composite Structures ballonets and ballast. Applied Mathematics and Mechanics 2005;26(8):
2006;75(1–4):151–5. 1072–82.
[73] Pagitz M, Pellegrino S. Shape optimization of ‘Pumpkin’ balloons. In: AIAA [105] Abramson HN. The dynamic behavior of liquids in moving containers with
balloon systems conference, Williamsburg, VA, May 21–24, 2007. applications to space vehicle technology. NASA SP-106; 1966.
[74] Maekawa S, Yoshino T. Tear propagation of a high-performance airship [106] Ibrahim RA. Liquid sloshing dynamics: theory and applications. Cambridge,
envelope material. Journal of Aircraft 2008;45(5):1546–53. UK: Cambridge University Press; 2005. p. 296–337.